Review on Superpixel Generation Algorithms Based on Clustering
Zhijie Chen, Baolong Guo, Cheng Li, Hongyan Liu
Abstract
Superpixel is often used in image preprocessing, which is generated by the feature similarity between pixels. Superpixel can capture redundant information of image and reduce dimension without sacrificing too much precision. These advantages make it receive more and more attention from researchers at home and abroad. In this paper, we summarize the methods of generating superpixel based on clustering iteration and hope to provide some reference for related researchers. On the basis of extensive literature research, this paper firstly introduces the research status of the superpixel generation algorithm, combines with the basic ideas, innovations, advantages and disadvantages of various methods, and focuses on the induction and discussion of the superpixel generation algorithm based on clustering. We then compare the current common superpixel generation algorithms based on clustering and provide the performance and analysis results on the data set. At last, we introduce the research direction of superpixel technology in the future.