SVM-Based Edge persistence filtering algorithm-SVM for edge-preserving Filtering

SVM-Based Edge persistence filtering algorithm-SVM for edge-preserving Filtering

Qingxiong Yang shengnan Wang Narendra Ahuja 1. Introduction: In this article, the author proposes a method similar to bilateral
Filtering. The normalized convolution kernel of bilateral filtering is determined by the distance between the center pixel and Its pixel and the gray-scale change range. The distance and gray-scale change range are usually Gaussian. In this article, the author converts the filtering problem into a vector-mapping (vector-mapping approximation) Problem and uses Support Vector Machine (SVM) to solve the problem. Each pixel is represented by a feature vector consisting of the corresponding spatial filtered response and their products. Apply SVM regression
Regression) the ing function Si learned via e-SVM regression using the features and the corresponding bilateral filtered values from the traing Images ). This article is the first article to propose a learning-based bilateral filtering algorithm, which is effective for high-contrast Gaussian. In addition, its computing complexity is irrelevant to the gray-scale change value (our method ).
Is valid for both low and high range variance Gaussian and the computational complexity is independent of the range variance value ). 2. Method Introduction-(because the formula is not easy to edit). For the Method Introduction, refer to the original article. Below are several images. 3. The paper also compares several other related algorithms (comparison ).