Since the correlation jam has a serious influence on the feature extraction of remote sensing images

as well as on the precise identification of the remote-sensing ground-objects

the paper not only presents small-and large-visual angle models

but also gives corresponding transform space Hough algorithm on the basis of the quantitive analysis of the amplitude and the phase. Compared with the methods of neighbour average

median filtering and classical frequency space filtering

the algorithm proposed in the paper is useful in removing the random-phase jam of the images

reserving richer fine textures and edge information and getting higher PSNRs. The algorithm has been applied successfully to the real time aquisition and processing system for remote sensing images.