毛叶枣的计算机视觉分级技术研究

Automatic Classification of Jujube Based on Computer Vision

根据农业部发布的农业行业标准NY/T484-2002,基于NI Vision Assistant系列软件对毛叶枣提出了一种自动分级方法。以黑色传送带为背景,采用CCD摄像机在毛叶枣样本的滚动中采集图像,通过对图像进行初步分割、灰度化、平滑去噪、增强、边缘检测、二值化等处理得出样本果实大小、表面缺陷大小以及果梗识别,进而借助机械手完成对毛叶枣品质的自动分级,识别结果与人工挑选结果吻合率分别达到:优等品吻合92.43%,一等品吻合96.34%,二等品吻合95.60%,残次品吻合95.95%。

According to the agriculture industry standard NY / T 484-2002 issued by the ministry of agriculture,this text supply an automatic classification method of Jujube based on a series of NI vision assistant software.Black belt as the background,using CCD cameras collect images when Jujube samples were rolling,getting the area of the surface defect and distinguish its stem from the image by operation as partition,graying,smooth filter,enhancement,sedge detection,two-value,etc.,and then finish the jujube automatic classification by mechanical hand.