A comparison of inter-frame feature measures for robust object classification in sector scan sonar image sequences

This paper presents an investigation of the robustness of an inter-frame fe ature measure classifier for underwater sector scan sonar image sequences. In the initial stages the images are of either divers or remotely operated vehicles (ROV's). The inter-frame feature measures are derived from sequenc es of sonar scans to characterize the behavior of the objects over time, Th e classifier has been shown to produce error rates of 0%-2% using real nonn oisy images. The investigation looks at the robustness of the classifier wi th increased noise conditions and changes in the filtering of the images. I t also identifies a set of features that are less susceptible to increased noise conditions and changes in the image filters. These features are the m ean variance, and the variance of the rate of change in time of the intra-f rame feature measures area, perimeter, compactness, maximum dimension and t he first and second invariant moments of the objects. It is shown how the p erformance of the classifier can be improved, Success rates of up to 100% w ere obtained for a classifier trained under normal noise conditions, signal -to-noise ratio (SNR) around 9.5 dB, and a noisy test sequence of SNR 7.6 d B.