基于内容的SIFT+LSH管道缺陷检索算法研究

Research on Content Based SIFT+LSH Pipeline Defect Retrieval Algorithm

各个城市对地下管道安全的检测一直是研究的热点和难点;传统的检测仪器不仅费时费力而且误检率特别高,随着技术的发展计算机视觉相关的方法也有在管道检测中应用,但是检测的速度和效果不尽人意;针对当前传统的检测方法操作复杂,成本高的问题,提出了一种基于内容的SIFT+LSH管道缺陷图像检索方法;该方法首先选取了优势较为明显的局部特征SIFT,充分利用了管道缺陷图像的特征,同时选取LSH算法对图像SIFT特征进行优化,将其转化为Hash编码,提高了检索速度;实验结果表明,基于SIFT特征和LSH算法的管道缺陷检索方法,相比与传统的SIFT特征和欧式距离的检索方法,大大提高了检索的速度,使得检测人员在实际操作中能够更快地获取到比较满意的检索结果。

The detection of underground pipeline safety in various cities has been a hot and difficult issue.Traditional inspection instruments are not only time-consuming and laborious,but also have a high rate of false detection.With the development of technology,computer vision related methods are also used in pipeline inspection,but the speed and effect of detection are not satisfactory.In view of the complexity and high cost of the traditional detection methods,a content based SIFT+LSH algorithm for pipeline defect image retrieval is proposed.This method first selects SIFT features more obvious advantages,make full use of the characteristics of pipeline defect image,select the LSH algorithm to optimize the image SIFT feature,it is transformed into Hash encoding,and improve the retrieval speed.The experimental results show that the retrieval method of pipeline SIFT features and LSH algorithm based on the defects,compared with the traditional SIFT feature retrieval method and Euclidean distance,the retrieval speed is greatly improved,so the detection personnel can quickly get satisfactory retrieval results in the actual operation.