Edge Information Detection of Remote Sensing Image Based on Two-dimensional Otsu Algorithm

摘要

Because of the problem that the conventional edge detection operator is difficult to effectively extract high-resolution remote-sensing image edge information, a high-resolution remote sensing image edge information extraction method based on two-dimensional Otsu algorithm is proposed. Firstly, used median filter algorithm to carry out denoising for original image, and used contrast enhancement method to improve the divisibility among ground objects of remote sensing image; secondly, used two-dimensional Otsu algorithm to carry out threshold segmentation for filtered image, obtained the binary image; finally, used a new adaptive Canny operator to detected the edge information for binary image, obtained edge information. In order to verify the feasibility and reliability of the proposed method, we selected two groups of Quickbird images for experiment. The experiment results show that compared with traditional Canny operator, edge information detection accuracy of the proposed method is higher.