Signal processing
This chapter has introduced basic signal processing techniques and has, by means of contributions, provided examples of some of the methods and principles of GPR processing being currently developed. The selection of suitable signal processing methods must start from a clear appre ciation of the modulation technique and the likely form of the received wavelet. The main initial objective is to select suitable processing to optimise the wavelet output in terms of each individual A-scan sample time series, and deconvolution techniques have been described. If the subsequent objective is to generate an image it is reason able to consider some type of 3D 'spiking' filter or migration of the data. If, however, the objective is to classify the wavelet, i.e. by Prony processing, then 'spiking' filters are not appropriate. Consideration can also be given to the removal of multiple reflec tions. Once the A-scan data are optimised, processing methods based on B-scan data sets can be considered. Again, if the objective is image 'spiking' or focusing, a num ber of migration or synthetic aperture methods are available, each of which is more or less tolerant to variations in propagation conditions. Examples of the application of migration techniques have been given. Where this method is not preferred, image pattern recognition techniques based on standard image processing methods of template matching can be used. Transforms such as Hough or neural network techniques can also be considered.
Signal processing, Page 1 of 2
< Previous page Next page > /docserver/preview/fulltext/books/ra/pbra015e/PBRA015E_ch7-1.gif /docserver/preview/fulltext/books/ra/pbra015e/PBRA015E_ch7-2.gif