半导体光电, 2019, 40 (6): 902, 网络出版: 2019-12-17
基于图像处理的暖血器缺陷检测方法研究
Research on Defect Detection Method of Blood Warmer Based on Image Processing
机器视觉 阈值分割 图像处理 缺陷识别 塑制品 machine vision threshold segmentation image processing defect identification plastic products
摘要
针对人工检测效率低、暖血器表面灰度值变化平缓以至缺陷提取受限的问题, 提出一种基于图像处理的暖血器缺陷检测方法。首先合理搭建视觉检测平台, 对采集图像进行灰度化、直方图均衡化等预处理; 其次改进阈值分割算法, 对灰度映射变换直方图进行拟合并引入判断条件, 对单峰灰度图直接进行自动阈值分割, 将双峰、多峰灰度图分解为多个部分并运用迭代求取最佳阈值T; 最后运用形态学提取侧面轮廓信息, 完成缺陷分类。实验结果表明, 该方法与常用方法相比, 能够有效区分缺陷种类, 并将准确率提升至99.33%, 满足企业实际检测要求。
Abstract
In view of the low efficiency of manual detection, the slow change of gray scale on the surface of the blood warmer and the limitation of defect extraction, a method of testing the defect of the heater based on image processing was proposed. First of all, a visual detection platform was set up to make pretreatments such as graying and histogram equalization on the collected images. Secondly, the threshold segmentation algorithm was improved, the gray mapping transformation histogram was fitted and judgment conditions were introduced to automatically threshold segment the single-peak gray map directly, the double-peak and multi-peak gray maps were decomposed into multiple parts, and the optimal threshold T was obtained by iteration. Finally, morphology was used to extract the profile information and complete the defect classification. Experimental results show that this method can effectively distinguish defect types and improve the accuracy to 99.33%, which meets the actual detection requirements of enterprises.
全燕鸣, 龙思慧, 吴航. 基于图像处理的暖血器缺陷检测方法研究[J]. 半导体光电, 2019, 40(6): 902. QUAN Yanming, LONG Sihui, WU Hang. Research on Defect Detection Method of Blood Warmer Based on Image Processing[J]. Semiconductor Optoelectronics, 2019, 40(6): 902.