Potato mechanical damage detection is a necessary step to realize the automatic classification of potato. In or-der to overcome the disadvantages of the artificial detection methods which are low efficiency and easy to be subjectively influenced, this paper proposes a novel method of machine vision detection based on high pass filter. First, images are collected according to the requirement of potato detection. Then, the H component image segmentation algorithm is used for the gray map of potatoes. Afterwards, we construct a Gauss high pass filter to obtain high frequency region by per-forming convolution with the fast Fourier transform of the gray image of the potato. Finally, the target area is obtained through the Blob analysis. Experimental results show that the method can accurately identify the mechanical damage of potato, the total recognition rate may reach 95%.%马铃薯机械损伤检测是实现马铃薯自动化分级的必要步骤。为了克服人工检测方法低效且易受主观因素影响的缺点，提出了一种基于高通滤波的机器视觉检测方法。首先，根据马铃薯检测的需求，采集图像；然后，利用 H 分量图像分割算法得到马铃薯灰度图；接着，构造高通滤波器，通过与快速傅立叶变换后的马铃薯灰度图做卷积，得到高频部分；最后，通过 Blob 分析筛选得到目标区域。实验结果表明：该方法能够较准确识别马铃薯机械损伤缺陷，总识别率达95%。
展开▼
