水果图像的背景分割和边缘检测技术研究

Study on background segment and edge detection of fruit image using machine vision

根据区域阈值法分别对所采集的彩色图像的红、绿、蓝三帧图像进行背景分割 ,发现在 B分量灰度直方图中利用双峰法选择阈值进行背景分割的效果最好 ;利用多种微分算子提取水果图像的阶跃性边缘 ,并用 Hilditch细线化方法对已提取边缘的图像进行了细线化处理 ,获得了较好的边缘提取和细化处理效果 ;所研究的背景分割和边缘检测技术的处理效果可以满足进一步进行水果的尺寸、果梗。

In the light of the characteristics of gray level variations in the image, bimodal gray region segment method was adopted to segment the pear from the background in the three images (red, green, blue). The best segment result was appeared in B gray level histogram. Four different edge operator were used to extract the step like edge of the fruit image, and the Hilditch thinning method was used to fulfil edge thinning. As the results shown, the adopted method of background segment and edge detection could satisfy the requirement of detecting size, shape, stem, and surface defect of fruit.