Circular shape detection in remote sensing multispectral images

In this paper a technique originally developed for computer vision, the Hough Transform, has been applied to remote sensing image analysis. A new implementation of the Hough Transform to detect circles which is suitable for the design of an automatic tool for remote sensing image analysis has been developed. Synthetic and real images have been analyzed in order to understand critical factors in circle detection. Results obtained both on some TV camera and LANDS AT test images showing circular features of geological interest are reported. A performance analysis has been done by first processing test images by means of the proposed Hough Transform methodology and then by comparing the obtained results with those supplied by photointerpretation.