基于矢量相似性的多元滤波方法研究.

In order to avoid the problem of sorting in color image filtering, this paper proposed a multivariate filter method for color image based on the study of vector similarity. Firstly, in RGB color space, it defined the similarity measure between the color vectors by the distance and the angle to describe the color vector similarity consistent with the human visual perception. Secondly, it designed and constructed the color multivariate filter using the color similarity criterion, and analyzed and studied in depth the influence of the related parameters on the filtering performance. Finally, in order to verify the effectiveness of the proposed method, it was applied to the actual remote sensing image filtering. It can be seen from the experiment that the proposed method not only effectively solves the problem of sorting of the traditional filtering methods, but also overcomes the problems of blurring and unclear edges due to the filtering. In addition, compared with the results of other traditional methods, the results show that the proposed method not only can effectively filter many kinds of noise, but also maintains the original image information better, making the image information clear and fidelity. The visual effect of this filtering method is superior to the traditional, its objective evaluation index also has greatly improved . [ABSTRACT FROM AUTHOR]