基于图像边缘形态学分析的轴承质检方法

Quality Inspection Method of Axletree Based on Image Edge Morphologic Analysis

提出一种在工业零件质量检测环境中判断轴承质量的图像识别方法,使用滤波、图像增强和分割等工序对图像进行预处理。给出相对方向编码的概念,对二值图像的边缘进行平滑处理。提出一种新的边缘形态学分析的方法对二值化图像边界形态进行量化分析,并运用神经元网络分类器对图像进行分类。实验结果表明,该方法能达到较好的识别效果。

This paper proposes a method of detecting industrial axletree quality based on image edge morphologic analysis and recognition technology. The image filter, enhancement and segmentation are adopted to pre-process the image and a novel image coding method based on relative direction coding is presented as well. Furthermore, an improved method based on image edge morphologic analysis is proposed, which can analyze the image edge with large quantities, a neural network classifier is utilized to sort images. Experimental results show that the approach to detecting axletree quality can achieve well recognition effect.