Accelerated superpixel image segmentation with a parallelized DBSCAN algorithm

作     者：Loke, Seng Cheong MacDonald, Bruce A. Parsons, Matthew Wunsche, Burkhard Claus 

作者机构：Univ Auckland Fac Med & Hlth Sci Auckland New Zealand Univ Auckland Fac Engn Auckland New Zealand Univ Waikato Hamilton New Zealand Univ Auckland Fac Sci Auckland New Zealand 

出 版 物：《JOURNAL OF REAL-TIME IMAGE PROCESSING》 (实时图像处理杂志)

年 卷 期：2021年第18卷第6期

页      面：2361-2376页

核心收录：

学科分类：0808[工学-电气工程] 1002[医学-临床医学] 08[工学] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：University of Auckland 

主　　题：Computational photography Concurrent algorithms DBSCAN Image segmentation Memory allocation Superpixels 

摘      要：Segmentation of an image into superpixel clusters is a necessary part of many imaging pathways. In this article, we describe a new routine for superpixel image segmentation (F-DBSCAN) based on the DBSCAN algorithm that is six times faster than previous existing methods, while being competitive in terms of segmentation quality and resistance to noise. The gains in speed are achieved through efficient parallelization of the cluster search process by limiting the size of each cluster thus enabling the processes to operate in parallel without duplicating search areas. Calculations are performed in large consolidated memory buffers which eliminate fragmentation and maximize memory cache hits thus improving performance. When tested on the Berkeley Segmentation Dataset, the average processing speed is 175 frames/s with a Boundary Recall of 0.797 and an Achievable Segmentation Accuracy of 0.944.