An improved SLIC superpixel algorithm based on nonlinear filtering

Abstract

Simple Linear Iterative Clustering (SLIC) algorithm is one of the best methods to generate superpixels in terms of segmentation effect and running speed. In an effort to solve the problem of inaccurate segmentation in traditional SLIC method, an improved SLIC algorithm is put forward in this paper. This algorithm adopts a nonlinear filter to preprocess the image, which can remove noise and smooth the image while maintaining the edge information. According to experimental results, the improved SLIC superpixel algorithm can generate satisfactory superpixels in visual segmentation effect and the boundary recall rate without too much time complexity.