基于色彩纹理特征的车门打磨维修检测

Door polishing service area inspection based on color texture characteristics

为解决自动化识别车门打磨维修区域检测的问题,设计了一种可用于车门打磨维修质量判断的图像预处理算法。首先,从色彩特征方面对比阈值分割各算法效果,确定区域生长法采取8邻域并设定阈值M为25实现区域划分。然后,从纹理特征方面分析,设计基于局部二值模式(LBP)的圆形LBP216法,并采用双线性插值法进行像素拟合,实现纹理提取。最后,融合上述算法作为图像预处理算法,实现二值图像的输出,并且提高优化算法时效性以及辨识精度,满足后续图像分类要求。

In order to solve the problem of automatic identification of door polishing and maintenance area detection,an image preprocessing algorithm that can be used to judge the quality of door polishing and maintenance was designed.Firstly,from the aspect of color characteristics,the effect of the threshold segmentation algorithm was compared,and the regional growth method was determined to adopt 8 neighborhoods and set the threshold M to 25 to achieve regional division.Then,from the aspect of texture features,a circular LBP216 method based on local binary mode(LBP)was designed,and the bilinear interpolation method was used for pixel fitting to achieve texture extraction.Finally,the above algorithm was used as an image preprocessing algorithm to realize the output of binary images,and improve the timeliness and recognition accuracy of the optimization algorithm to meet the identification requirements of image classification.