Edge detection applied to satellite imagery of the oceans

摘要

A computer edge-detection algorithm for automatic delineation of mesoscale structure in digital satellite IR (infrared) images of the ocean is developed. The popular derivative-based edge operators are shown to be too sensitive to edge fine-structure and to weak gradients to be useful in this application. The edge-detection algorithm is based on the gray level cooccurrence matrix (GLC), which is commonly used in image texture analysis. The cluster shade texture measure derived from the GLC matrix is found to be an excellent edge detector that exhibits the characteristic of fine-structure rejection while retaining edge sharpness. This characteristic is highly desirable for analyzing oceanographic satellite images.