An edge detection algorithm based upon the adaptive multi-directional anisotropic gaussian filter and its applications

Edge detection is essential to comprehend and analyze high-resolution remote sensing images. The Gaussian filter is widely used for image preprocessing prior to edge detection. However, some weak edge points in the images with low gradient values are more likely to be missed during edge detection following Gaussian filtering. Moreover, balancing the demands of denoising and sharpening edges is challenging in the Gaussian filter. An edge detection algorithm based on the adaptive multi-directional anisotropic Gaussian filter (AMDAGF) is proposed in this article to address these issues. Meanwhile, this article proposes an effective solution for the adaptive setting of the key parameters of the proposed algorithm, which provides assistance for the application and promotion of the proposed algorithm. Finally, by using ROC curve analysis technology, comparative edge detection experiments about two experimental areas were performed, and the superiority and viability of the proposed algorithm have been demonstrated.