Detection and classification of circular structures on spot images

The unsupervised method proposed here for detecting structural features on satellite images consists of three major steps: a. The extraction of contours, which depends on the encountered texture, is obtained by an iterative filtering followed by several thresholds that generate binary images. The thresholds are determined by different percentages of pixels over the total number of pixels contained in the image. We then trace the limits of the binary forms previously filtered by an iterative majority smoothing. b. The detection of the structures from the contours involves four substeps: individualization of the curves; decomposition of the curves into subcircular elements; application of a version of the Hough transform to each subcircular elements; computation of precise results. c. The computation of parameters that discriminate the detected structures; namely, for each structure, the position of the center and radius of the reference circle, the number of detected pixels, the chord and the intersection coefficients, the normal direction and the distance between the middle of the chord, and the intersection point between normal and chord. This set of data allows us to select the different families of structures that we are looking for. As an example, the method has been applied to the region of Azru, which presents on both geological and geomorphological levels, numerous circular structures of varied origins.