Yu.G. Sosulin, A.B. Russkin
In the paper the questions of development and performance analysis of detection extensive low-contrast targets algorithm in underlying surface images with using fractal dimension are considered. The block diagram of fractal target detector and basic parameters are determined. The procedure for detection threshold determining to provide a given level of false alarms probability is proposed. The comparative analysis of twelve fractal dimension estimation methods based on a stochastic fractal reflected surface model with respect to estimation accuracy and time complexity is considered. The performance evaluation and comparative analysis of fractal detectors of extensive low-contrast targets, which used different fractal dimension estimation methods, from simulation procedure are fulfilled. Investigations on the fractal processing of real SAR images and empirical detection characteristics definition of artificial objects on the underlying surface backgrounds based on radar data analysis are carried out. The performance evaluation and comparative analysis of fractal detectors depending on fractal dimension estimation method are executed. The possibility of extensive low-contrast targets detection on the fractal surface background on real data with given detection probability and false alarms rate is determined. The comparison of the results obtained on simulation procedure and real radar images analysis is fulfilled.