This study presents an overview of various three-dimensional (3D) image steganography techniques from survey point of view. The authors present taxonomy of 3D image steganography techniques and identify the recent advances in this field. Steganalysis and attacks on 3D image steganography algorithms have also been studied. 3D image steganography techniques in all the three domains: geometrical, topological and representation domains have been studied and compared among each other on various parameters such as embedding capacity, reversibility and response towards attacks. Some challenges which inhibit the development of 3D steganography algorithms have been identified. This study concludes with some useful findings in the end. A comprehensive survey on 3D image steganography techniques, to the best of the authors’ knowledge, is not available and thus it suffices the need of this study.

The growth of the number of people in the monitoring scene may increase the probability of security threat, which makes crowd counting more and more important. Most of the existing approaches estimate the number of pedestrians within one frame, which results in inconsistent predictions in terms of time. This study, for the first time, introduces a quadratic programming (QP) model with the network flow constraints to improve the accuracy of crowd counting. Firstly, the foreground of each frame is segmented into groups, each of which contains several pedestrians. Then, a regression-based map is developed in accordance with the relationship between low-level features of each group and the number of people in it. Secondly, a directed graph is constructed to simulate constraints on people's flow, whose vertices represent groups of each frame and arcs represent people moving from one group to another. Finally, by solving a QP problem with network flow constraints in the directed graph, the authors obtain consistency in people counting. The experimental results show that the proposed method can reduce the crowd counting errors and improve the accuracy. Moreover, this method can also be applied to any ultramodern group-based regression counting approach to get improvements.

Preserving details while restoring images highly corrupted by impulsive salt and pepper noise remains a challenging problem. The authors proposed an algorithm based on radial basis functions (RBFs) interpolation which estimates the intensities of corrupted pixels by their neighbours. In this algorithm, intensity values of noisy pixels in the corrupted image are first estimated using RBFs. Next, the image is smoothed. The proposed algorithm can effectively remove the highly dense, impulsive salt and pepper noise. Experimental results show the superiority of the proposed algorithm both in noise suppression and details preservation in comparison to the recent similar methods. Extensive simulations show better results measured by peak signal-to-noise ratio and structural similarity index, especially when the image is corrupted by very highly dense impulse noise.

In this study, the authors examine the issue of numerical accuracy of different variants of integral image computation algorithms both sequential and parallel (e.g. well-suited for graphics processing units). The experimental study shows that particular arrangements of successive arithmetical operations for parallel integral image computation algorithms as well as a row–column approach can lead up to a substantial numerical accuracy improvement over classic implementations. Such accuracy differences are then experimentally shown, on the example of the image binarisation problem, to be of great significance in terms of quality of final results being obtained by image processing methods utilising integral images as a part of their processing framework. In addition, the authors draw conclusions which can later be used in the formulation of general guidelines for constructing high-quality and time-effective implementations of image processing methods involving integral images.

This study presents a chaos-based crypto-compression scheme to protect the H.264 advanced video coding (AVC) used for the applications of video conference. To this end, a selective encryption approach was adopted. The authors proposed to encrypt the inter-prediction, the intra-prediction and the context adaptive variable length coding. The format compliance and compression ratio are conserved. Their encryption method is based on two piecewise linear chaotic maps for pseudo-random bit generators. The first is considered as a decision module to choose either to encrypt or not, whereas the second is used for the encryption. The proposed encryption techniques were developed and discussed, and their experimental results indicate that the scheme is secure and very efficient according to the computing times. Moreover, it is suitable for real-time application and convenient for all the H.264/AVC profiles or even with other video compression standards.

Optical coherence tomography is an immersive technique for depth analysis of retinal layers. Automatic choroid layer segmentation is a challenging task because of the low contrast inputs. Existing methodologies carried choroid layer segmentation manually or semi-automatically. The authors proposed automated choroid layer segmentation based on normalised cut algorithm, which aims at extracting the global impression of images and treats the segmentation as a graph partitioning problem. Due to the structure complexity of retinal and choroid layers, the authors employed a series of pre-processing to make the cut more deterministic and accurate. The proposed method divided the image into several patches and ran the normalised cut algorithm on every patch separately. The aim was to avoid insignificant vertical cuts and focus on horizontal cutting. After processing every patch, the authors acquired a global cut on the original image by combining all the patches. Later the authors measured the choroidal thickness which is highly helpful in the diagnosis of several retinal diseases. The results were computed on a total of 525 images of 21 real patients. Experimental results showed that the mean relative error rate of the proposed method was around 0.4 when compared with the manual segmentation performed by the experts.

For high-resolution synthetic aperture radar (SAR), constant false alarm rate (CFAR) detectors are widely used to separate targets from background and it is proved that generalised gamma distribution () can model the non-homogeneous clutter appropriately. However, CFAR detectors based on possess two problems. First, the methods for solving highly non-linear equations of resort to numerical iterative algorithms, which are computationally expensive. To avoid this, the authors present a novel analytic solution for parameter estimation of by exploiting a third-order approximation of the polygamma function. This novel analytic solution can result in more accurate parameter estimation and can fit a wider range of the parameters. The second problem is that in a multi-target SAR image, some target pixels may be classified as clutter pixels. To select pixels of interest, an iterative sliding window approach is often used in CFAR. They analyse the relationship between the detection probability and the number of iterations, and prove that this strategy can reduce the miss rate and false detection rate effectively. On the basis of the aforementioned parameter estimation method and the iterative sliding window approach, an innovative and effective CFAR detector is proposed in this study and its superiority is demonstrated by experiments.

To further improve the effect of infrared small target detection, a reweighted infrared patch-image model is proposed. First, the authors point out that the nuclear norm in the infrared patch-image model could easily leave some sparse background edges in the target patch-image, leading to an inaccurate background estimation. Then, to overcome this defect, the reweighted nuclear norm is adopted to constrain the background patch-image, which could preserve the background edges better. Considering that some non-target sparse points could not be suppressed by only using l ₁ norm, the authors introduce the reweighted l ₁ norm to further enhance the sparsity of target image. Finally, the proposed model is formulated as a reweighted robust principal component analysis problem and solved by the inexact augmented Lagrangian multiplier method. Extensive experiments show that the proposed model outperforms the other six competitive methods in suppressing background clutter and detecting target.

Recognition of complex static hand gestures is a challenging problem due to the complexity of hand gestures, which are rich in diversities because of high degrees of freedom involved by the human hand. It is more difficult especially when gestures are represented by two hands. This study proposes a framework that can recognise complex static hand gestures by using the wristband-based contour features (WBCFs). The authors require the user to wear a pair of black wristbands on his (her) two hand wrists, so that the hand region(s) can be segmented accurately. The topmost and sharpest corner point of the wristband on a gesturing hand is detected first. It is treated as a landmark to extract the WBCF of a hand gesture. Then, a simple feature matching method is proposed to obtain a recognition result. To deal with the cases where hand region(s) cannot be segmented correctly, watershed segmentation, and region merging techniques are adopted to provide improvements on hand region segmentation. Experimental results show that their system can be used to recognise 29 Turkish fingerspelling sign hand gestures and achieve a recognition accuracy of 99.31% with only six training images for each gesture.

Haze removal is a non-trivial work for unmanned aerial vehicle (UAV) aerial video processing, and challenges are mainly attributed to spatial-temporal coherence and computational efficiency. The authors propose a novel dehazing algorithm for hazy UAV aerial video and improve the classical dark channel prior approach with a bright region filling process to alleviate colour distortion in the recovered video, which enhances the spatial consistency. To achieve better temporal coherence, the authors constrain the atmospheric light estimation between adjacent frames by using a temporal filter. The authors also optimise the transmission calculation to reduce computational complexity. Experimental findings show that the proposed algorithm yields results superior to those obtained from previous methods. Compared with that in frame-by-frame dehazing, the processing time in the proposed method is reduced by 74.5%.

Motion vectors (MVs) are the most common temporal descriptors in video analysis, indexing and retrieval applications. However, video indexing and analysis based on MVs do not perform well for videos at different dimension ratios (DRs) or even various resolutions. As a result, video indexing and analysis which are based on identifying similar video face with many difficulties at different DRs or resolutions by MVs. In this study, a two-stage algorithm is introduced to make MV descriptors robust against variations first in DR and then at resolution. In the experiments performed on motion vector histograms, the proposed method improves the performance on identifying similar videos at various spatial specifications by up to 73%. Moreover, in the video retrieval experiments, the proposed modified MV outperforms original MV feature vector. This is an indication of improvement in differentiation of similar and dissimilar videos by the proposed temporal feature vector.

Speckle noise is one of the major artefacts in ultrasound images. The denoising faces the trade-off between noise suppression and structural preservation. In this study, multiscale adaptive regularisation Savitzky–Golay (MARSG) method, the new filter for removing speckle noise, is proposed. The proposed method combines the benefit of the multiscale analysis and the outstanding noise removing capability of Savitzky–Golay (SG) filter. The Laplacian pyramid is employed to separate an image into the noise, texture and object layers. Adaptive regularisation Savitzky–Golay (ARSG) filter is developed as the denoising filter in the noise and the texture layers. The denoising of the ARSG filter is adaptively adjusted in order to preserve the edges of objects in the image. The experiments on the synthetic and ultrasound images demonstrated that MARSG method offered better balance between noise removal and structural preservation than non-linear multiscale wavelet diffusion, feature-enhanced speckle reduction and regularised SG filter.

The recent development of advanced television systems has demonstrated a need for an efficient video conversion technique. In this scenario, frame rate up conversion (FRUC) solutions play an important role due to their benefits in both increasing the viewing quality experience and reducing the cost of video transmission. However, with the recent increase in video resolution, notably from standard definition to high definition (HD) and ultra HD, FRUC now requires not only better interpolated frame quality but also lower FRUC time processing. Considering this problem, this study proposes a novel statistical learning based adaptive search range (SR) solution to enable an effective FRUC mechanism. In the proposed adaptive SR solution, a set of spatial-temporal features are carefully defined and exploited to adaptively assign an appropriate SR value to each considered block, notably by formulating the SR adaptation as a classification problem and using the well-known support vector machine framework for the classification task. Experimental results conducted on a rich set of common video test sequences show the advantages of the proposed adaptive SR solution, notably in both interpolated frame quality improvement and time processing reduction.

The influence of environmental light sources affects the colour cast in natural images. In computer vision, biased colours have a significant influence on object recognition and classification. Illuminant estimation aims to eliminate these effects and obtain the image in canonical white light. In this study, the authors propose a deep non-negative matrix factorisation (DeepNMF) method to estimate the illuminant of colour-biased images. DeepNMF deeply factorises the input matrix into multiple layers, separating the image into patches and reshaping each channel of the patch as an [R,G,B] matrix. Based on the diagonal model, they assume that the final layer is the estimated illuminant of each patch. Mean pooling is then used to estimate the illuminant of the overall image. The angular error is used as a metric to test the authors’ method on three commonly used colour constancy datasets. The results show that the proposed method is comparable to state-of-the-art methods, although it is simpler to implement. As the proposed method uses a single image as input, it does not require a learning process.

To achieve quick and accurate segmentations for intensity inhomogeneous images, the model named fractional B-spline level set (FBSLS) is proposed in this study. The core of FBSLS method is that the LS function is expressed as the line combination of FBS basis in continuous forms. In this expression, the evolution of the energy function can be seen as a variation problem on the space spanned by the FBSs. As a result, the minimum value of energy function can be obtained directly from the coefficient of FBSs. Furthermore, every minimisation step can be considered as a convolution operation, and the evolution of the process of energy function may be regarded as the filtering operation whose kernel function is defined by the fractional order BS function. The filtering operation induces an intrinsic algorithm of smoothing which may be controlled explicitly by the order of the chosen FBS function. At the last, the morphological operator was used as the step of post-processing to get better segmentation. The obtained experimental results demonstrate that the proposed model has better segment results than the traditional ones on segmentation time, Dice coefficient, precision, and recall and F-measure metrics.

A novel approach to detect and localise anomalous events in crowed scenes by processing surveillance videos is introduced in this study. Unusual events are those that significantly differ from current dominated behaviours. The proposed approach both detects pixel-level and block-level anomalies. In pixel level, Gaussian mixture models are used to detect abnormalities. Block-level detection segments the crowd into blocks according to pedestrian detection, and then anomalies are spotted and localised with a social force model. Experimental results using the USCD datasets Ped1 and Ped2 show that the proposed method performs favourably against state-of-the-art methods.

A novel fusion algorithm based on guided filter (GF) for computed tomography (CT) and magnetic resonance imaging (MRI) medical images is proposed. In this algorithm: approximation coefficient and three wavelet coefficients of CT and MRI are obtained by the wavelet transform, respectively. Two weight maps are obtained by comparison of the pixel values of the two approximation coefficients. A GF is designed with the weight maps serving as the input image and the corresponding approximation coefficient serving as the guided image; the GF is used to smooth the weight images and refined weight maps are obtained. The approximation and wavelet coefficients of CT and MRI images are fused by the weighted fusion algorithm with refined weight maps. A fused image of CT and MRI is obtained by the inverse wavelet transform. Comparisons of this algorithm with two fusion algorithms available show that the fused image based on this algorithm contains a greater amount of information, more details and clearer edges than the other two algorithms. Therefore, this algorithm is better at locating the position and shape of the target volume. In the course of treatment, this algorithm can better avoid the surrounding health organs by radiation, protect the health of patients.

Fingerprint enhancement plays a very important role in automatic fingerprint identification system. In order to ensure reliable fingerprint identification and improve fingerprint ridge structure, a novel method based on the collaborative filtering model for fingerprint enhancement is proposed. The proposed method consists of two stages. First, the original fingerprint is pre-enhanced by using Gabor filter and linear contrast stretching. Next, the pre-enhanced fingerprint is partitioned into patches in spatial domain, and then the patches are enhanced based on spectra diffusion by using the two-dimensional (2D) angular-pass filter and the 2D Butterworth band-pass filter. The proposed method takes full advantage of the ridge information and spectra diffusion with higher quality to recover the lost ridge information. To evaluate proposed method, the databases FVC2004 are employed, and the comparison experiments are carried out using various methods. Comparative experimental results show that the proposed algorithm outperforms the existing state-of-the-art methods on fingerprint enhancement.

In the literature, different types of algorithms that are organised to encrypt and decrypt images have been introduced. Some of these depend on chaotic systems where bifurcation routes to chaos exist. Those algorithms have advantages and disadvantages so far as their security level and computational speed are concerned. This study proposes a robust algorithm based on a pixel shuffling and a one-dimensional chaotic economic map for encrypting and decrypting images. The proposed algorithm is implemented on many images. The security and performance of the proposed method are analysed thoroughly by using key space, key-sensitivity, correlation of two adjacent pixels, information entropy, contrast and differential attack. On the basis of the obtained experimental results, the proposed algorithm is characterised by a large size of key space, a high sensitivity to the secret key, very low correlation coefficients, a good information entropy and a high contrast. Finally, the experiments are confirmed that the proposed algorithm can resist statistical and differential attacks with high efficiency.