GUO Zongjian, DENG Shaoqiang, TANG Kezong

(School of Information Engineering, Jingdezhen Ceramic University, Jingdezhen 333403, Jiangxi, China)

Abstract: The traditional Otsu algorithm needs to traverse each gray value of an image and that computational complexity isincreased sharply with increasing image quantization level. A new adaptive particle swarm optimization (APSO) algorithm isproposed for image segmentation. According to the particle fitness, in this APSO algorithm, a new inertia weight method wasintegrated for adjusting the particle flight speed and a new extreme value perturbation method was constructed, which perturbsthe individual extreme value and the global optimal extreme value of the particle, so as to avoid the particle population to fall into the local optimal region. After simulating a group of ceramic and standard images with complex texture, the segmentationtime was averagely 215 ms less than that required in Otsu. For standard test images, the segmentation time was 82 ms less than that of Otsu. Our APSO algorithm can converge to the global optimal value in a short time and is more effective to segment thedetails of ceramic images with complex texture.
Key words: particle swarm optimization; image segmentation; Otsu algorithm; inertia weight