摘要
提出了一种基于形态学开闭运算和梯度优化的分水岭算法的目标检测方法。该方法首先利用形态学开闭运算对原始图像进行平滑处理,再对梯度图像进行阈值优化,去除过多的区域极小值,然后利用分水岭分割算法检测目标,最后利用目标的面积和空间关系等特征去除少量误提目标。实验表明,新方法可以取得很好的效果。
This paper proposed a method for object detection based on morphological opening-and-closing operation and gradi- ent optimization. Firstly, applied morphological opening-and-closing operation to smooth the original image. Secondly, in or- der to avoid overmuch segmented region, used threshold-value processing to gradient image before being segmented. Thirdly, applied watershed algorithm to detect object from SPOT image. At last, used object character such as area and space relation- ship to remove a few wrong objects from segmented results. Experimental results show that the new method may gain very good performance.
出处
《计算机应用研究》
CSCD
北大核心
2009年第4期1593-1594,1600,共3页
Application Research of Computers
基金
国家自然科学基金资助项目(60665001)
关键词
形态学开闭运算
梯度优化
分水岭
目标检测
morphological opening-and-closing operation
gradient optimization
watershed
object detection