摘要
提出一种基于人眼视觉识别的2次阈值分割和形态学消噪相结合的图像边界提取方法。针对光斑背景不稳定、亮度非高斯分布的特点,辅助以减背景图象预处理,消除背景变化对光斑质心位置的影响,依次用灰度和灰度对比度进行阈值分割,增强算法的实时性,并用矩形模板执行形态学开启和闭合操作,去除光斑内部无关的细节,获取连续的图像边界。该方法全天候对多个906 nm激光管产生的条带状激光光斑中心实时定位,时效性和定位精度均获得满意的效果,重复测量误差2%.
To detect a center of laser spots formed by many laser tubes at an open environment, two threshold division and morphologic filter were considered based on the identifying principle of human eye vision, and a novel method of edge extraction of laser spot was presented. The influence of background brightness variation on facular center position was avoided in terms of image pre-processing of decreasing background. An algorithm was operated more rapidly by use of grey-scale and grey-scale contrast as a threshold to divide laser image. Then the morphologic operators with a square model are executed to detect the image edge so as to eliminate the inner details and retain the edge continuity. This method was used to detect 906 nm laser spot center in all weather, and the experimental results show that the precision of detecting center of laser spot in real time is satisfied, and the error of replicate measurement is 2 %
出处
《兵工学报》
EI
CAS
CSCD
北大核心
2006年第5期823-826,共4页
Acta Armamentarii
关键词
光学
图像处理
灰度对比度
形态学
阈值分割
边界提取
optics
image processing
grey-scale contrast
morphology
threshold division
edge extraction
