Detection, estimation of directional features on ocean surface images

The detection-estimation of directional properties is proposed in the context of oceanographic remote sensing images collected by visible or radar sensors such as Spot and SAR Seasat. A complex natural scene can be viewed as composed of varied textured areas. However the amount of data and the unknown number of types of textures does not always allow exhaustive classification. Therefore, prior to an estimation of characteristic features, the authors propose a detection phase based on the recognition of anisotropic area. This step is decisive because knowledge of some characteristics of the image strongly influences the selection of possible estimated features. For wave pattern study emphasis is then given to pattern periodicity, length of wave crest and dominant direction field