基于传感器阵列的纹理图像表面缺陷识别算法  

Surface Defect Recognition Algorithm of Texture Image Based on Sensor Array

作　　者：滕碧红 孙海信[2] TENG Bi-hong;SUN Hai-xin(School of Mathematics and Computer Sciences,Fuzhou University,Fuzhou Fujian 350108,China;School of Information,Xiamen University,Fujian Xiamen 361000,China)

机构地区：[1]福州大学数学与计算机科学学院,福建福州350108 [2]厦门大学信息学院,福建厦门361000

出　　处：《计算机仿真》2023年第3期285-288,301,共5页Computer Simulation

基　　金：国家自然科学项目基金(62071402)。

摘　　要：针对目前方法对纹理图像进行缺陷识别时,由于未能利用结构方差直方图对纹理图像进行全局的特征提取,导致图像在进行缺陷识别时的小波系数收敛特性差、覆盖性能低、识别效果差的问题,提出基于传感器阵列的纹理图像表面缺陷识别算法。该算法首先将传感器阵列与小波去噪算法相结合对纹理图像进行去噪处理;利用灰度分布法将图像中的像素进行区间划分,通过结构方差直方图对图像的全局特征进行提取;再采用卡方距离对纹理图像像素之间的距离进行计算,最后通过最近邻分类器对计算结果进行分类,从而实现对纹理图像的表面缺陷进行识别。实验结果表明,运用上述算法对缺陷进行识别时的小波系数收敛特性好、覆盖性能高、识别效果好。Generally,during the defect recognition of texture image,the lack of global features of texture image extracted by structural variance histogram results in poor convergence characteristics of wavelet coefficients and recognition effect.Therefore,the surface defect recognition algorithm of texture image based on sensor array is studied in this paper.Firstly,the sensor array and wavelet denoising algorithm were combined to denoise the texture image.Then,the gray distribution method was introduced to divide the pixels in the image,extracting the global features of the image through the structural variance histogram.Secondly,the chi-square distance was utilized to calculate the distance between pixels of the texture image.Finally,according to the nearest neighbor classifier,the calculation results were classified to identify the surface defects of the texture image.The experimental results show that the algorithm has excellent wavelet coefficient convergence characteristics,high coverage performance and excellent recognition effect.

关 键 词：传感器阵列 纹理图像 缺陷识别 图像去噪 最近邻分类器 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]