基于机器视觉的图像边缘检测算法的优化研究

摘要

A method is proposed in this paper to resolve some problems in traditional edge detection ways.Problems,like inefficiency,low precision and low resolution of edge detection,appear in actual operation,which has affected the precision in the machine vision measurement.This paper proposes an adaptable methods and it consists of two main techniques;one of them is based on color difference and it is usual y applied in the occasion when the color gap is obvious;another method is the based on threshold segmentation that is always used when the color difference is inconspicuous.%提出了一种边缘检测算法，有效地解决了传统边缘检测算法在实际应用中的的效率不高、精度不够、不能适合实际问题的复杂性。边缘检测属于机器视觉测量中较为重要的部分，直接影响到测量的精度。文中提出的一种自适应边缘检测算法主要基于两种技术，一种技术基于颜色区分，通常用于颜色区分较明显的状态下；另一种基于阈值分割，常常用于颜色边界较为模糊的情况。通常情况下，阈值分割对光照阴影的敏感性较高，但是对颜色区分的影响较小；所以光照条件较差的条件下，基于颜色的边缘检测的方法能够很好地弥补阈值检测的缺点。颜色恒常性算法用来区分阴影的存在性。最后，在智能电表自动化质量检测系统的特征区域定位进行试验，精度和效率有显著的提高。