摘要
针对背景复杂、对比度低的红外舰船目标分割问题,提出了一种红外舰船图像分割的新算法.由于二维最大类间方差法不仅反映了图像的像素点灰度分布信息,还反映了邻域空间相关信息,因此有较好的抗噪能力.但是由于其解空间维数的增加,计算量的变化是以指数增长的,而粒子群优化算法可实现高效并行、随机、自适应群体搜索.基于这一特点,提出了基于粒子群优化的二维最大类间方差局部递归分割方法,有利于实现红外图像的实时处理.该方法同样适用于复杂背景下的其他红外目标图像的分割.
A novel infrared image segmentation algorithm for realizing infrared ship segmentation in the lower contrast and complicated background was presented. 2-D Otsu method not only considers the distribution of the gray information, but also takes advantage of the spatial neighbor information by using the 2-D histogram of the image, so it often gets better antinoise performance. However, its time-consuming computation is often an obstacle in application. Particle swarm optimization (PSO) algorithm can realize parallel, random and self-adapt colony search, hence an algorithm for PSO-based local recursive 2-D Otsu segmentation was proposed here. This algorithm can also be used in other infrared image segmentations with complicated backgrounds.
出处
《红外与毫米波学报》
SCIE
EI
CAS
CSCD
北大核心
2006年第4期295-300,共6页
Journal of Infrared and Millimeter Waves
基金
国家自然科学基金重点(F60135020)
国家重点预研(413010701-3)资助项目