New Publications are available for Image recognition
http://dl-live.theiet.org
New Publications are available now online for this publication.
Please follow the links to view the publication.Automatic identification of wildlife using local binary patterns
http://dl-live.theiet.org/content/conferences/10.1049/cp.2012.0454
Recognition of individuals is necessary for accurate estimation of wildlife population dynamics for effective management and conservation. Identifying individual wildlife by their distinctive body marks is one of the least invasive methods available. Although widely practiced, this method is mostly manual where newly captured images are compared with those in the library of previously captured images. The ability to do so automatically using computer vision techniques can improve speed and accuracy, facilitate on-field matching, and so on. This paper reports the results of using a texture based image feature descriptor, the Local Binary Patterns (LBP), for the automatic identification of an important endangered species - The Great Crested Newt (GCN). The proposed approach is tested on a database of newts' distinctive belly images which are treated as a source of biometric information. Results indicate that when both appearance and spatial information of newt belly patterns are encoded into a composite LBP feature vector, the discriminating power of the system can improve significantly. (6 pages)Illumination robust face representation based on intrinsic geometrical information
http://dl-live.theiet.org/content/conferences/10.1049/cp.2012.0437
The ability to recognize people is a key element for improving naturalistic human-robot and human-computer interaction systems. In this paper, we propose a binary non-subsampled contourlet transform (B-NSCT) based illumination robust face representation. Faces are transformed into multi-scale and multi-directional contour information where the intrinsic geometrical structures are used for characterising facial texture. Experiments on the Yale B and CMU PIE databases illustrate that B-NSCT is highly insensitive to illumination variation. (6 pages)Multi-frame super resolution using edge directed interpolation and complex wavelet transform
http://dl-live.theiet.org/content/conferences/10.1049/cp.2012.0447
In this paper, a multi frame super resolution technique is proposed which uses edge directional interpolation (EDI) and dual-tree complex wavelet transform (DT-CWT). In the proposed technique a super resolution process is applied for each frame to generate the low frequency component. On the other hand, high frequency components are generated by DTCWT decomposition followed by EDI. Finally, the composition of the generated subbands using inverse DTCWT (IDT-CWT) reconstructs the super resolved output frame. Experimental results on a number of benchmark video sequences with respect to their PSNR measures confirm the superiority of the suggested method over the state of the art video resolution enhancement methods. (5 pages)Geolocation: maps, measurements and methods
http://dl-live.theiet.org/content/conferences/10.1049/cp.2012.0404
The paper presents the geolocation is the art of innovative combination of properties of nature and sensor technology. The particle filter provides a general method for geolocation. Geolocation fingerprinting is a general concept for fitting a measurement (along a trajectory) to a geographical information system. Geolocation is equal to map plus measurement plus method. (48 pages)Preliminary evaluation of multispectral iris imagery
http://dl-live.theiet.org/content/conferences/10.1049/cp.2012.0450
The unique patterns of complex texture that are visible in iris images captured under near-infrared illumination are highly stable and can be used for high confidence biometric recognition. Similar patterns are visible in lighter colored irises under visible light. This paper presents an analysis of a database of iris images captured using multispectral illumination, from 405 nm to 1070 nm in wavelength. The analysis is based on matching performance using a Daugman-based commercial implementation of the iris recognition algorithm. We find that illumination wavelength has a very significant effect on iris recognition performance. (5 pages)Novel fingerprint segmentation with entropy-Li MCET using log-normal distribution
http://dl-live.theiet.org/content/conferences/10.1049/cp.2012.0455
Fingerprint recognition is an important biometric application. This process consists of several phases including fingerprint segmentation. This paper proposes a new method for fingerprint segmentation using a modified Iterative Minimum Cross Entropy Thresholding (MCET) method. The main idea is to model fingerprint images as a mixture of two Log-normal distributions. The proposed method was applied on bi-modal fingerprint images and promising experimental results were obtained. Evaluation of the resulting segmented fingerprint images shows that the proposed method yields better estimation of the optimal threshold than does the same MCET method with Gamma and Gaussian distributions. (6 pages)Face recognition using kernel collaborative representation and multiscale local binary patterns
http://dl-live.theiet.org/content/conferences/10.1049/cp.2012.0457
Collaborative Representation with regularized least square (CRC-RLS) is state-of-the-art face recognition method that exploits the role of collaboration between classes in representing the query sample. However, this method views the image as a point in a feature space, and the performance can be degraded when the cropped face image is misaligned and/or the lighting conditions change. Histogram-based features, such as Local Binary Patterns (LBP) have gained reputation as powerful and attractive texture descriptors showing excellent results in terms of accuracy in face recognition. In this paper, LBP features are introduced in CRC-RLS to confront these problems such as illumination. In addition, motivated by the recent success of non-linear approaches, a new kernel-based nonlinear regularized least square classifier with collaborative representation (KCRC-RLS) is proposed in this paper. The proposed system is evaluated on two benchmarks: ORL and Extended Yale B. The results indicate a significant increase in the performance when compared with state-of-the-art face recognition methods. (4 pages)Expression classification of 3D faces using local deformations
http://dl-live.theiet.org/content/conferences/10.1049/cp.2012.0452
One of the most important challenges for face recognition algorithms is dealing with large variability due to facial expression. This paper presents an approach for the expression classification of 3D face scans. The proposed method is based on modelling local deformations which are calculated as the surface change between a neutral face and a face with expression. These deformations are used to train a multiclass/multi-feature LDA classifier. On an unseen face local deformations are calculated automatically using a face with neutral expression as a reference. It is shown that the results obtained are comparable with other similar approaches with the advantage that there is not manual intervention is required for the classification process. (6 pages)Skin detection using color processing mechanism inspired by the visual system
http://dl-live.theiet.org/content/conferences/10.1049/cp.2012.0459
Skin detection is an important procedure for face detection, hand gesture analysis, and objectionable image filtering. In this paper, a skin detection algorithm is proposed according to the colour processing mechanism inspired by the visual system. In the algorithm different types of neurons are used to deal with skin colour and texture. Inspired by receptive fields of rod, cone and orientation selectivity neurons in the visual system, a neural network is proposed to identify skin area in an image. The network contains three types of cones, one type of rod and four types of orientation neurons. Three cones are in response to red, green and blue lights respectively. The rods are in response to brightness. Orientation neurons are used to detect textures of skin. Through their specific receptive fields, the network can be trained to identify skin areas. The experimental results show that the proposed algorithm is comparable to current skin segmentation algorithms. (5 pages)Feasibility of iris identification algorithm optimization by fractional template matching
http://dl-live.theiet.org/content/conferences/10.1049/cp.2012.0453
The human iris exhibits random and unique textural patterns that allow for identification with high accuracy. These patterns are evident in near-infrared (NIR) imagery, even for very dark irises. Iris templates are created from NIR iris imagery with the Ridge Energy Direction (RED) recognition algorithm and subsequently matched to measure performance. In this paper we investigate the feasibility of algorithm optimization by performing an initial fractional template comparison to eliminate high Hamming distance (HD) score matches followed by a full template re-comparison or iterative higher fractional template comparison on remaining templates. Recognition performance for different fractional template areas is analyzed with a view towards substantial improvement of algorithm identification time performance. The feasibility of identification time reduction by 60% or more is reported both ICE and Bath data sets. (5 pages)Optimal configuration strategies for iris recognition processing
http://dl-live.theiet.org/content/conferences/10.1049/cp.2012.0451
Although biometric systems are now widely used in a variety of scenarios, optimising practical systems can still be challenging because this can involve balancing a number of very different factors across a number of processing stages. In iris recognition, a typical processing chain leading from raw data capture to an identity decision requires several clearly identifiable modules, but within this chain it is generally accepted that the segmentation stage and the recognition module are particularly critical in determining overall performance. This paper investigates the extent to which overall performance can be optimised with respect to scenario constraints and requirements by an appropriate balance of processing power allocated between these two most important steps, and how enhanced processing at one step can compensate for less sophistication at another. The work reported thus provides some insight into practical options for system implementation which can take account of constraints in, for example, error rate performance, processing speed, resource allocation, and so on. (6 pages)Advanced video camera identification using conditional probability features
http://dl-live.theiet.org/content/conferences/10.1049/cp.2012.0426
Today, the misuse of digital data especially images and videos become crucial with the existence of sophisticated high-tech equipment and it is available at relatively low cost. Illegal recording of movie in cinema has caused losses of millions of dollars a year. Law enforcement agencies are keen to find ways to counter illegal video recording. Current research into camera identification techniques is attracting a significant amount of attention. The main objective is to identify the camera equipment used to record digital image or video based on the data source obtained. In this paper, we propose a video camera identification technique based on the Conditional Probability (CP) Features. Specifically we focus on its performance for identification of video sources using cameras of different models. In our experiments, we demonstrate that the CP Features are able to correctly match the test video frames with their source with classification accuracy is approximately 97.2%. These findings provide a good indication that CP Features are suitable for digital video forensics. (5 pages)Geometric feature based age classification using facial images
http://dl-live.theiet.org/content/conferences/10.1049/cp.2012.0438
This paper presents the use of geometric feature based models for age group determination of facial color images. This process consists of two main stages: geometric feature extraction, analysis and age group classification. The feature extraction was performed with the correct understanding of the effect of age on facial anthropometry. The age differentiation capability of the features is evaluated using three different classifiers, namely, neural network classifier, support vector classifier, normal densities-based linear classifier. The facial face images are categorized to five major age groups. To show the effectiveness and accuracy of the proposed feature extraction, experiments are conducted on two publically available databases namely FGNET and IFDB. The results show that the success rate of classification is around 90%. (5 pages)Model, predict and test: towards a rigorous process for acquisition of object detection and recognition algorithms for un-manned air vehicle applications
http://dl-live.theiet.org/content/conferences/10.1049/cp.2012.0423
The designers of unmanned air vehicle (UAV) mission systems are seeking to exploit advances in consumer image processing technology, to provide additional object detection and recognition capability for UAV systems. However, the two application domains are quite different, so a simple transfer of algorithms is not possible. A formal approach for selecting and developing algorithms is proposed. This involves algorithm modelling and experimental validation by the algorithm supplier, together with assessment trials in a representative system, carried out by the system designer. The objective is to build confidence in the behaviour of the algorithm under consideration, to build high fidelity models that can be used in system design studies and to address integration issues early in a development programme. A review of the open literature indicates that the necessary algorithm modelling is feasible and an example is presented to illustrate the proposed approach. (6 pages)Retinal vessel segmentation using ensemble classifier of bagged decision trees
http://dl-live.theiet.org/content/conferences/10.1049/cp.2012.0458
This paper presents a new supervised method for segmentation of blood vessels in retinal images. This method uses an ensemble system of boot strapped decision trees and utilizes a feature vector based on the orientation analysis of gradient vector field, morphological linear transformation, line strength measures and Gabor filter responses. The feature vector encodes information to handle the healthy as well as the pathological retinal image. The method is evaluated on the publicly available DRIVE and STARE databases. Method performance on both sets of test images is better than the 2<sup xmlns="http://pub2web.metastore.ingenta.com/ns/">nd</sup> human observer and other existing methodologies available in the literature. The incurred accuracy, speed, robustness and simplicity make the algorithm a suitable tool for automated retinal image analysis. (6 pages)Object search using wavelet-based polar matching for aerial imagery
http://dl-live.theiet.org/content/conferences/10.1049/ic.2011.0167
A method for finding known objects in aerial and satellite imagery was designed, implemented in MATLAB and tested. This method took a keypoint-based approach to describing objects. Several of the steps used in the scheme took advantage of the Dual-Tree Complex Wavelet Transform (DTCWT). These were combined with a new approach which takes the highly localised information encoded in keypoint descriptors and combines it with “global” geometrical information about the target. Many keypoints which corroborate the same approximate location as the predicted target centre can improve match confidence. The implementation of the above was evaluated both with synthetic imagery, as a means to identify and characterise factors which degrade performance, and also real imagery as a means to demonstrate its “real-world” performance. The real imagery demonstrated success in identifying a range of objects. (5 pages)Randomised forests for people detection
http://dl-live.theiet.org/content/conferences/10.1049/ic.2011.0148
People detection is an important task with applications in fields such as surveillance and human computer interaction. A popular approach to this problem is to train a classifier on a data set using a particular set of features. A great deal of empirical evidence suggests that edge features are particularly discriminative for this task. In this paper we explore the use of randomised forests (sometimes referred to as randomised decision forests) for people detection. A randomised forest classifier is trained for people detection with edge orientation features. These features capture information concerning the distribution of edges with specific orientations. The classifier is trained and tested on the INRIA person data set, and some results are presented. (5 pages)Facial blood vessels activity in drunk persons using thermal infrared
http://dl-live.theiet.org/content/conferences/10.1049/ic.2011.0108
In this work, the activity of the facial blood vessels of drunk persons is studied using thermal infrared images. Nonlinear anisotropic diffusion is applied to enhance the vessels on the images and after that top-hat transformation is used for isolating them from the rest information on the face. Simple thresholding is applied to raise the more active vessels. Registration procedures are employed to easily compare the vessel activity on the face of the drunk person with that on the sober person. In drunk persons, vessels around nose and eyes as well as on the forehead become more active. This work constitutes a preliminary study, which aims at qualitative and not quantitative results. Basically, the restricted number of the 20 persons that participated in the experiment is not considered adequate for statistical inference. (5 pages)Real time car theft decline system using ARM processor
http://dl-live.theiet.org/content/conferences/10.1049/ic.2011.0059
Due to the insecure environment the ratio of vehicle theft increases rapidly. Because of this is manufacturers of luxury automobiles has the responsibilities for taking steps to ensure the authorization for the owners and also inbuilt the anti theft system to prevent the car from theft. The existing system was. Car alarm techniques are used to prevent the car theft with the help of different type of sensors like pressure, tilt and shock & door sensors.Drawbacks are cost and cant used to find out the thief, it just prevents the vehicles from loss. The proposed security system for smart cars used to prevent them from loss or theft using Advanced RISC Machine (ARM) processor. It performs the real time user authentication (driver, who starts the car engine) using face recognition, using the Principle Component Analysis - Linear Discreminant Analysis (PCA LDA) algorithm. According to the comparison result (authentic or not), ARM processor triggers certain actions. If the result is not authentic means ARM produces the signal to block the car access (i.e. Produce the interrupt signal to car engine to stop its action) and inform the car owner about the unauthorized access via Multimedia Message Services (MMS) with the help of GSM modem. Also it can be extends to send the current location of the vehicle using the GPS modem as a Short Message Services (SMS) as passive method.Improving real time video surveillance performance using inter-frame retransmission
http://dl-live.theiet.org/content/conferences/10.1049/ic.2011.0102
Transmission Control Protocol (TCP) and User Datagram Protocol (UDP) are common transport layer protocols used in the IP based networks. However, both protocols have disadvantages for wireless video transmission. TCP yields high delay, while UDP experiences high packet loss. Lost packet retransmission is one of the solutions in video transmission to enhance video quality. This paper proposes inter-frame retransmission method to enhance the performance of video surveillance over WiMAX. The method outperforms the existing retransmission protocols. (5 pages)Efficient face recognition algorithms based on transformed shape features
http://dl-live.theiet.org/content/conferences/10.1049/ic.2011.0123
Human face recognition is, indeed, a challenging task, especially under the illumination and pose variations. It is challenging as well as attractive for its usefulness in the area of crime detection and identity verification. We examine in the present paper effectiveness of two simple algorithms using coiflet packet and Radon transforms to recognize human faces from some databases of still graylevel images, under the environment of illumination and pose variations. Both the algorithms convert 2-D graylevel training face images into their respective depth maps or physical shape which are subsequently transformed by Coiflet packet and Radon transforms to compute energy for feature extraction. Experiments show that such transformed shape features are robust to illumination and pose variations. With the features extracted, training classes are optimally separated through linear discriminant analysis (LDA), while classification for test face images is made through a k-NN classifier, based on L<sub xmlns="http://pub2web.metastore.ingenta.com/ns/">1</sub> norm and Mahalanobis distance measures. Proposed algorithms are then tested on face images that differ in illumination, expression or pose separately, obtained from three data bases, namely, ORL, Yale and Essex-Grimace databases. Results, so obtained, are compared with two different existing algorithms. Performance using Daubechies wavelets is also examined. It is seen that the proposed Coiflet packet and Radon transform based algorithms have significant performance, especially under different illumination conditions and pose variation. Comparison shows the proposed algorithms are superior. (6 pages)Iris-biometric comparators: minimizing trade-offs costs between computational performance and recognition accuracy
http://dl-live.theiet.org/content/conferences/10.1049/ic.2011.0110
The intricate structure of the iris constitutes a powerful biometric utilized by iris recognition algorithms to extract discriminative biometric templates. In order to provide a rapid comparison of biometric templates the vast majority of feature extraction methods are designed to generate binary biometric templates, applying the Hamming distance as (dis-)similarity metric. Based on this concept several feature extraction techniques have been proposed in literature, while potential improvements in comparison procedures are commonly neglected. In this paper trade-off costs between the computational performance and recognition accuracy of iris-biometric comparators are investigated. Different comparison techniques of binary biometric templates, and a composition of these, are proposed, where emphasis is put on the trade-off between computational cost and improvement of recognition accuracy, i.e. recognition accuracy is improved at minimal additional computational cost. Experimental results confirm the soundness of the proposed approaches. (6 pages)An object tracking in particle filtering and data association framework, using SIFT features
http://dl-live.theiet.org/content/conferences/10.1049/ic.2011.0104
In this paper, we propose a novel approach for multi-object tracking for video surveillance with a single static camera using particle filtering and data association. The proposed method allows for real-time tracking and deals with the most important challenges: (1) selecting and tracking real objects of interest in noisy environments and (2) managing occlusion. We will consider tracker inputs from classic motion detection (based on background subtraction and clustering). Particle filtering has proven very successful for non-linear and non-Gaussian estimation problems. This article presents SIFT feature tracking in a particle filtering and data association framework. The performance of the proposed algorithm is evaluated on sequences from ETISEO, CAVIAR, PETS2001 and VS-PETS2003 datasets in order to show the improvements relative to the current state-of-the-art. (6 pages)Audio event detection and localisation for directing video surveillance - a survey of potential techniques
http://dl-live.theiet.org/content/conferences/10.1049/ic.2011.0118
Hearing is often a vital adjunct to vision for situational awareness. This article discusses how automatic techniques for audio event detection and localisation can bring major benefits when incorporated in video surveillance systems. There is an overview of the key technological components, with a particular focus on microphone array techniques for sound source localisation, and how techniques originally developed for voice communications can be adapted to surveillance applications. (4 pages)Image CAPTCHA based on distorted faces
http://dl-live.theiet.org/content/conferences/10.1049/ic.2011.0106
An image recognition-based CAPTCHA is proposed for increasing security in web applications. The proposed method uses distorted faces to create an image for a CAPTCHA test. The user has to recognise the well-known person that appears in the image choosing the name from a list. The method uses a feature-line morphing technique to distort the faces which morphs the well-known person's face into a cartoon or an animal. The performance of this approach is evaluated through different face recognition systems. The results show an improvement in human recognition in comparison with word-based CAPTCHAs and an increment in robustness against robots when trying to break through the tests. (6 pages)Minutiae + friction ridges = triplet-based features for determining sufficiency in fingerprints
http://dl-live.theiet.org/content/conferences/10.1049/ic.2011.0099
In order to provide statistical and qualitative backing to latent fingerprint evidence, an algorithm is proposed to discover statistically rare features or patterns in fingerprint images. These features would help establish an objective minimum- quality baseline for latent prints as well as aid in the latent examination process in reaching a matching decision. The proposed algorithm uses minutia triplet-based features in a hierarchical fashion, where minutia points are used along with ridge information toestablish relations between minutiae. Preliminary results show that a set of distinctive features can be found that have sufficient discriminatory power to aid in quality assessment. An example set of 10 statistically rare features is presented, resulting from analysis of a set of 93 images. (6 pages)Multi-vehicle convoy analysis based on ANPR data
http://dl-live.theiet.org/content/conferences/10.1049/ic.2011.0135
This paper focuses on the development and novel application of data mining techniques for convoy analysis of vehicles based on the automatic number plate recognition (ANPR) system. The amount of ANPR data captured daily by traffic cameras in the road networks is very substantial. Data mining techniques are commonly used to extract relevant information and to reduce the amount of data processing and storage. In this paper, we apply data clustering techniques to extract relevant traffic patterns from the ANPR data to detect and identify unusual patterns and irregular behaviour of multi-vehicle convoy activities. (5 pages)Detecting object in the dynamic background from the noisy image in visual surveillance
http://dl-live.theiet.org/content/conferences/10.1049/ic.2011.0062
Detecting an object from a dynamic background is a challenging process m computer vision and pattern matching research. The proposed algorithm identifies moving objects from the sequence of video frames which contains dynamically changing backgrounds in the noisy environment. In connection with our previous work, here we have proposed a methodology to perform background subtraction and modernized from moving vehicles in traffic video sequences that combines statistical assumptions of moving objects using the previous frames in the dynamically varying noisy situation. For that, a binary moving objects hypothesis mask is constructed. Then, Kalman filter is utilized for the amalgamation of current background. Shadow and noise removal algorithms are proposed to operate at the lattice which identifies object-level elements. The results of post-processing can be used to detect object more efficiently. Experimental results and comparisons using real data demonstrate the pre-eminence of the proposed approach.Evaluating iris segmentation for scenario optimisation
http://dl-live.theiet.org/content/conferences/10.1049/ic.2011.0098
Iris recognition is a biometric modality which offers the potential for high accuracy and, increasingly, for application in more diverse environments than hitherto. Poor segmentation is one of the most important factors likely to compromise iris recognition performance. Hence, research in the area of iris biometrics has often been focused on efforts to enhance the performance of iris segmentation techniques, and this has led to considerable work on iris segmentation. This paper presents a detailed investigation, evaluation and comparison of several segmentation approaches (including a new algorithm proposed by the authors) proposed in the literature based on their accuracy and processing speed. To be consistent with the research of others, for all quantitative experiments, algorithms have been evaluated on two iris databases, namely CASIA V1.0 and a subset of the BioSecure database. (6 pages)Self-dependent 3D face rotational alignment using the nose region
http://dl-live.theiet.org/content/conferences/10.1049/ic.2011.0101
One of the challenging issues for 3D face recognition is face alignment. Many alignment algorithms are computationally expensive, making them unsuitable for real-time biometrics, or not robust enough to detect large variations in pose. In this work, a novel algorithm for 3D face rotational alignment is proposed, that uses the nose region. After preprocessing and nose region identification, alignment is performed by applying two energy functions to the nose footprint, identified as the largest filled region in the inverted depth map. These functions are minimised using Simulated Annealing and the Levenberg-Marqurdt algorithm. The energy minimisation and segmentation procedures continue iteratively until a stopping criterion is met. The method has been applied to images from the Face Recognition Grand Challenge (FRGC) v2 dataset and the consistency of its alignment has been verified using the iterative closest point (ICP) algorithm. As a self-dependent algorithm, it does not require a pre-aligned image as a reference and also has a high computational speed, approximately three times faster than the brute force ICP technique. (6 pages)Hostile intent and behaviour detection in elevators
http://dl-live.theiet.org/content/conferences/10.1049/ic.2011.0115
We propose a visual surveillance based person-to-person hostile intent and behavior detection method in elevators. The view of an elevator by a surveillance camera is typically of a small confined space with abrupt changes in illumination due to opening and closing of the elevator door. We extract three levels of features in a sequential process for the violent event detection. First, as low-level features, foreground blobs are segmented from the background and their motion velocity vectors are extracted by an optical flow method. Second, as a mid-level feature, the number of people inside the elevator is estimated by considering the number and sizes of the segmented blobs. As the other mid-level features, the velocity magnitudes and directions are computed by image based motion analyses. A person-to-person violence can only occur when there is more than one person in the elevator. As the key classifying feature, we consider the average velocity magnitude and direction of each blob. A sequence of image frames are determined to contain a violent event if an average velocity magnitude of any segmented blob exceeds a threshold along with its associated direction not being dominant in one direction. The experimental results demonstrate that the proposed method functions effectively with a computational efficiency sufficient for real-time processing. (6 pages)Color face recognition using quaternion PCA
http://dl-live.theiet.org/content/conferences/10.1049/ic.2011.0109
Recently, biometric systems have attracted the attention of both academic and industrial communities. Advances in hardware and software technologies have paved the way to such growing interest. Nowadays, efficient and cost-effective biometric solutions are continuously emerging. Fingerprint-based biometric systems have emerged as pioneering commercial applications of biometric systems. Face and iris traits have proven to be reliable candidates. Until recently, face recognition research literally followed the research undertaken in the field of fingerprint recognition which is inherently gray-scale. In this paper, efforts are restricted to the investigation of face representations in the color domain. The concept of principal component analysis (PCA) is carried over into the hypercomplex domain (i.e., quaternionic) to define quaternionic PCA (Q-PCA) where color faces are compactly represented. Unlike the existing approaches for handling the color information, the proposed algorithm implicitly accounts for the correlation that exists between the face color components (i.e., red, green and blue, respectively). (6 pages)Human gait recognition: approaches, datasets and challenges
http://dl-live.theiet.org/content/conferences/10.1049/ic.2011.0100
As a new technology of biometrics, gait recognition has attracted a great deal of interest in computer vision community due to its advantage of unobtrusive recognition at a distance. In this paper, a review of this topic is presented in terms of approaches, databases and challenges. First, a comprehensive survey of recent developments on gait recognition approaches is reported, where relevant techniques are categorized into three classes for discussions. Then, more than 10 publically available datasets are analysed and compared in detail. In addition, challenges that constrain practical application of gait recognition systems are discussed. Finally, future trends are given to guide further research in this field. (6 pages)Strip shredded document reconstruction using optical character recognition
http://dl-live.theiet.org/content/conferences/10.1049/ic.2011.0132
Document reconstruction affects different areas such as archeology, philology and forensics. A reconstruction of fragmented writing materials allows to retrieve and to analyze the lost content. Due to the complexity of reconstruction, automated algorithms are necessary. A reconstruction methodology for shredded documents is presented in this paper which recognizes characters at the stripes' borders and matches them subsequently. In order to achieve this, an Optical Character Recognition (OCR) system is exploited, that is capable of recognizing partially visible characters by means of local features. Thus, no binarization needs to be performed. Preliminary results show the ability of matching shredded documents using the information of cut characters. (6 pages)Object classification based on behaviour patterns
http://dl-live.theiet.org/content/conferences/10.1049/ic.2011.0112
With the recent explosion of surveillance videos, media management has gained n increasing popularity. Addressing this challenge, in this paper, we propose a Surveillance Media Management framework for object detection and classification based on behaviour patterns. The objectives of the paper are: (i) demostrating the discriminative power of behaviour features for object recognition and classification, (ii) proposing a behavioural fuzzy classifier which progressively discriminate objects by including different degrees of uncertainty in the classification process and (iii) presenting a Surveillance Media Management system to extract semantic media information and provide unsupervised object classification from raw surveillance videos. The performance of the proposed system has been thoroughly evaluated on AVSS 2007 surveillance dataset and as the results indicate the proposed technique enhances object classification performance. (6 pages)Iterative active querying for surveillance data retrieval in crime detection and forensics
http://dl-live.theiet.org/content/conferences/10.1049/ic.2011.0133
Large sets of visual data are now available both, in real time and off line, at time of investigation in multimedia forensics, however passive querying systems often encounter difficulties in retrieving significant results. In this paper we propose an iterative active querying system for video surveillance and forensic applications based on the continuous interaction between the user and the system. The positive and negative user feedbacks are exploited as the input of a graph based transductive procedure for iteratively refining the initial query results. Experiments are shown using people trajectories and people appearance as distance metrics. (6 pages)Quality-assured fingerprint image enhancement and extraction using hyperspectral imaging
http://dl-live.theiet.org/content/conferences/10.1049/ic.2011.0120
Hyperspectral imaging is an emerging technology with a wide range of potential applications including remote sensing, chemical and pharmaceutical, food quality control and forensic science. In this paper, fingerprint extraction from various surfaces using visible and near-infrared hyperspectral imaging for forensic applications is explored. Firstly, spectral analysis is conducted, where principal component analysis (PCA) is employed to determine the most informative linear band combinations of the hypercube. Then, based on the fingerprint quality index measurement, the optimal principal component is determined to have the highest quality score. Accordingly, the fingerprint image is also enhanced due to the effect of PCA. Comprehensive experimental results are reported and analysed. (6 pages)Estimation of 3D head region using gait motion for surveillance video
http://dl-live.theiet.org/content/conferences/10.1049/ic.2011.0105
Detecting and recognizing people is important in surveillance. Many detection approaches use local information, such as pattern and colour, which can lead to constraints on application such as changes in illumination, low resolution, and camera view point. In this paper we propose a novel method for estimating the 3D head region based on analysing the gait motion derived from the video provided by a single camera. Generally, when a person walks there is known head movement in the vertical direction, regardless of the walking direction. Using this characteristic the gait period is detected using wavelet decomposition and the heel strike position is calculated in 3D space. Then, a 3D gait trajectory model is constructed by non-linear optimization. We evaluate our new approach using the CAVIAR database and show that we can indeed determine the head region to good effect. The contributions of this research include the first use of detecting a face region by using human gait and which has fewer application constraints than many previous approaches. (6 pages)Consistent quality control for wireless video surveillance using distributed video coding
http://dl-live.theiet.org/content/conferences/10.1049/ic.2011.0116
Distributed Video Coding (DVC) is well known for low complexity encoding which provides coding solutions for a wide range of applications, in particular wireless video surveillance. In this paper, we address the problem of distortion variation introduced by typical rate control algorithms, especially in a various bit rate environment. A distortion quantization model derived from a MPEG-2 distortion estimation model is proposed to achieve smooth picture quality across video frames. Simulation results show that the proposed quality control algorithm is capable to meet user defined target distortion and maintain a rather small variation for sequence with slow motion and performs similar to fixed quantization for fast motion sequence at the cost of some RD performance. (6 pages)Biometric authentication for secured transaction using finger vein technology
http://dl-live.theiet.org/content/conferences/10.1049/cp.2011.0465
In this Paper, Finger vein authentication is used to ensure the user identity. Finger vein technology, is one of the fastest growing method in the Biometrics. Vein patterns are different for each and ever person also they are hidden under the skins surface. Since they are situated inside forgery is extremely difficult. Near IR rays are passed through the finger and the hemoglobin pigments inside the blood blocked the rays and the patterns of vein are captured at the other side via Image sensor. HausdThorff distance technique is used to extract the Skelton pattern from the vein image. This paper discusses the utilization of Minutiae feature points in the verification tasks as geometric representation of vein pattern shape. Also discusses the technology and applications of finger vein authentication, as well as the importance of authentication in its future development.An improved method using kinematic features for action recognition
http://dl-live.theiet.org/content/conferences/10.1049/cp.2011.0766
Human action recognition is a challenge problem in computer vision. In this paper, we propose an improved approach using kinematic features for action recognition. In this approach, we find the area that relates to action by a simple method, and select eight discriminative features derived from optical flow field to describe the dynamics of the field. The covariance matrix of the feature vectors is used to fuse the features and to serve as the feature descriptor. Multi-class SVM classifiers are then employed for action classification. Experiments are carried out on public datasets. We obtain a recognition rate of 97.66% SEG-ACA and 98.2% SEQ-ACA on KTH dataset, and 98.89% SEQ-ACA and 93.83% SEG-ACA on WEIZMANN dataset with leave-one-out test.Automatic authorized vehicle recognition system
http://dl-live.theiet.org/content/conferences/10.1049/cp.2011.0471
The growth of technology is increasing day by day to fulfill the human needs. The proposed system is implemented to make human work easier. ANPR is an image processing technology which uses number plate to identify the vehicle. The objective is to design an efficient automatic authorized vehicle identification system by using the vehicle number plate. The system is implemented on the entrance of a highly restricted areas for security e.g. Parliament, military zones, Supreme Court etc. The proposed algorithm consists of four major parts: vehicle detection, extraction of plate region, classification of a vehicle and recognition of plate characters. For vehicle detection moving object detecting algorithm is used. For extracting the plate region smearing algorithm and segmentation are used. The alpha numeric characters are recognized by using OCR. The resulting data is then used to compare with the database so as to come up with the specific information like the vehicle's owner, place of registration, address, etc. The system is implemented and simulated in Mat lab, and its performance is tested on a real image.Design of identity verification unit and management system
http://dl-live.theiet.org/content/conferences/10.1049/cp.2011.0777
With an increasing emphasis on the emerging automatic handicap personal identification applications, living fingerprint-based identification and Smart cards are receiving a lot of attention. So as to achieve secure and reliable identity verification and the tracking of documentary files and stationary, the paper presents an identity verification unit and management system based on living fingerprint identification techniques and Internet of things (IOT). Additionally, for the goal of non-paper office, the unit resolves issues, e.g., the insecurity and simplex function in traditional identification unit, by providing professional function such as electronic seal.Facial expression recognition based on local facial regions
http://dl-live.theiet.org/content/conferences/10.1049/cp.2011.1002
In facial expression recognition, the extraction of expression features is most important. It is a complex problem that how to extract steady and effective expression features from facial images which contain much disturbance. lt is the key to improve recognition rate, too. In this paper, four local regions such as mouth, nose, eye (including the eyebrows) and glabella that are extracted. Our experiments are performed on the database of CED-WYU (1.0). And the experiment results explain our understanding about facial expression recognition based on local facial regions.Object detection in smoke screen interference image sequences based on fractal
http://dl-live.theiet.org/content/conferences/10.1049/cp.2011.0994
In the paper, a new method for the detection of moving weak targets in smoke screen interference image sequences is proposed. Edges in infrared image add a deterministic element to surface height degrading locally the surface 'fractality' and leading to an underestimation of the local fractal dimension (LFD). A different local fractal dimension will be determined at edges even if the two segments incident to the edge have the same LFD. The concept is evaluated by comparing row-mean subtraction filter based on fractal theory with conventional operators such as, median subtraction filter. Results show a similar performance in a low-noise environment and superiority of the fractal operators in a high noise, the algorithms are effectively for smoke screen interference and are easy to be implemented by parallel processing hardware.Mobile augmented reality system for personal museum tour guide applications
http://dl-live.theiet.org/content/conferences/10.1049/cp.2011.0887
In this paper, we developed a prototype of a mobile interactive museum guide system, which consists of an ultra mobile PC equipped with a webcam. This museum guidance system can automatically find and retrieve multimedia information about the objects of interest to the visitors in an intuitive way. A coarse to fine image recognition method is used to improve the recognition rate and a sub-exhibits localization method is proposed to solve the occlusion problem.A general HCI system based on monocular camera for intelligent terminals
http://dl-live.theiet.org/content/conferences/10.1049/cp.2011.1489
General human-computer interaction (HCI) based on monocular camera is one of the most important areas where rapid and comprehensive intelligence of network terminals and natural interaction systems can be achieved using image sensors. Currently, sensor networks require lots of expensive and single function sensors, and at the same time there exists some complexity of interaction between human beings and objects as well as objects and objects. Under these circumstances, the development of the intelligent network terminals based on image sensors is becoming increasingly significant. In this paper, we propose an interaction system based on HCI technology achieved by the most widely used image sensors, such as PC cameras, phone cameras and surveillance cameras in public places. The system applies simple but fast image processing technologies, such as image segmentation, tracking, recognition and some well-improved methods.Regularized neighborhood boundary discriminant analysis for facial expression recognition
http://dl-live.theiet.org/content/conferences/10.1049/cp.2011.0999
In this paper, we propose an approach which named regularized neighborhood boundary discriminant analysis for facial expression recognition. Our algorithm is based on the linear boundary discriminant analysis (LBDA), which aims to find a optimal projection in order to enhance the ability of classification. A regularized method was executed to remove the singularity of within-class metric matrix. Experiments on JAFEE facial expression database and Cohn-Kanade database show that our proposed method can get better performance than some other methods, such as linear discriminant analysis (LDA), local fisher discriminant analysis (LFDA) and linear boundary discriminant analysis (LBDA).Facial expression recognition using local binary covariance matrices
http://dl-live.theiet.org/content/conferences/10.1049/cp.2011.0997
In this paper, we propose a novel local feature descriptor for facial expression recognition, referred as local binary covariance matrices (LBCM). The covariance matrix in LBCM is constructed by incorporating location, intensity and local binary features of each pixel inside a region of interest. Extensive experiments demonstrate the effectiveness of LBCM for FER, even if partial occlusion exists.A fall detection algorithm based on pattern recognition and human posture analysis
http://dl-live.theiet.org/content/conferences/10.1049/cp.2011.0790
Detecting fall is a particular important task in security monitoring and healthcare applications of sensor networks. However traditional approaches suffer from either a high false positive rate or high false negative rate, especially when the collected sensor data are unbalanced. Therefore, there is a lack of tradeoff between false alarms and misses for many traditional data mining methods to be applied. To solve this problem a novel fall detection algorithm based on pattern recognition and human posture analysis is presented in this paper. It firstly extracts thirty temporal features from the original data traces for different length adaptation of samples, and then exploits Hidden Markov Model (HMM) to filter the noisy character data and reduce the dimension of feature vectors. After that, it performs a closer classification with one-class Support Vector Machine (OCSVM) to filter the high false positive samples, and finally applies posture analysis to counteract the effects of high false negative samples until a satisfying accuracy is achieved. Simulation with real data demonstrates that the proposed algorithm outperforms other existing approaches.