基于小波兴趣突出点的破损轮胎图像检索方法-《计算机技术与发展》

文章信息/Info

Title:: Image Retrieval Method of Damaged Tire Based on Wavelet Interest Salient Points

作者:: 张素智¹; 常俊²; 吴玉红²; 1. 郑州轻工业大学软件学院,河南郑州 450007;
2. 郑州轻工业大学计算机与通信工程学院,河南郑州 450007

Author(s):: ZHANG Su-zhi1; CHANG Jun2; WU Yu-hong2; 1. School of Software,Zhengzhou University of Light Industry,Zhengzhou 450007,China;
2. School of Computer and Communication Engineering,Zhengzhou University of Light Industry,Zhengzhou 450007,China

Keywords:: damaged tire image retrieval; wavelet salient points of interest; feature extraction; circular color histogram; binarization

摘要:: 简单叙述了破损轮胎图像检索的背景意义,描述了图像检索研究的发展及现状,分析了基于人工和兴趣点的图像检索方法的不足,提出了一种基于小波兴趣突出点的破损轮胎图像检索方法。该方法需检测破损轮胎图像的兴趣突出点,具体做法是在小波域提取兴趣突出点,将小波兴趣突出点标记在轮胎图像中破损区域并提取特征,用该特征来描述破损轮胎图像;为了弥补传统颜色直方图仅有色彩信息而缺少其颜色信息的相关位置信息描述的不足,设计了环形二值化颜色直方图。在提取破损轮胎图像特征时,利用环形二值化颜色直方图对小波兴趣突出点的颜色特征和其相对位置特征相结合的方法进行特征提取,图像检索时的小波兴趣突出点邻域内的二值化颜色直方图就增加了位置描述的功能,使得提高破损轮胎图像检索准确率成为了可能。实验结果表明,该方法对破损轮胎图像的检索是有效可行的。

Abstract:: We briefly describe the background and significance of image retrieval of damaged tire,discuss the development and status quoof image retrieval research,analyze the shortcomings of image retrieval methods based on artificial and interest points,and propose amethod of damaged tire image retrieval based on wavelet interest salient points. This method needs to detect the salient points of interestin the damaged tire image. The specific method is to extract the salient points of interest in the wavelet domain,mark the salient points ofinterest in the damaged area of the tire image and extract the features,and use the features to describe the damaged tire image. In order tomake up for the deficiency of the traditional color histogram which only has color information but lacks the description of the relevant location information of its color information,a circular binary color histogram is proposed. When extracting the features of the damaged tireimage,the circular binary color histogram is used to extract the features of the color features and the relative position features of the salientpoints of the wavelet interest. The binary color histogram in the neighborhood of the salient points of the wavelet interest in image retrieval adds the function of position description,which makes it possible to improve the accuracy of the damaged tire image retrieval. Experiment shows that the proposed method is effective and feasible for image retrieval of damaged tire.