LED芯片图像的势函数标记分水岭分割*

作者：李 蒙，方家乐，朱颖合 日期：2010-09-08/span> 浏览：3700 查看PDF文档

LED芯片图像的势函数标记分水岭分割*

李 蒙，方家乐，朱颖合

(杭州电子科技大学 自动化学院,浙江 杭州 310018)

摘要：全自动LED芯片测试分拣系统通过视觉图像处理精确定位LED芯片，其图像特征信息的提取建立在良好的图像分割基础上。为了精确提取图像特征，针对LED芯片图像的特点，改进了分水岭分割算法，采用直方图势函数提取标记，并在标记基础上对梯度图进行分水岭变换，实现了LED芯片图像的良好分割。实验结果表明，该方法有效地抑制了过分割现象，具有较好的抗噪性，对LED芯片图像感兴趣区域提取准确度高、鲁棒性强，分割效果较好。
关键词：LED芯片图像；势函数；标记；分水岭
中图分类号：TP391.41
文献标识码：A文章编号：1001-4551(2010)07-0081-04

Markercontrolled watershed segmentation of potential function for LED die image
LI Meng, FANG Jiale, ZHU Yinghe
(School of Automation, Hangzhou Dianzi University, Hangzhou 310018, China)

Abstract: Accurate location by computer vision and image processing is used in automatic LED die testing and sorting system. The feature extraction used to calculate and measure the parameters of LED die is based on the quality of image segmentation. In order to extract the image features accurately, an effective markercontrolled watershed segmentation method was presented according the features of LED die image. Histogram potential function was involved in order to get better markers which made watershed segmentation better. The results show that the proposed method could segment the LED die image well. The phenomena of oversegmentation is well restrained, and robust to noise.
Key words: LED die image; potential function; marker; watershed