Research Of Instrument Detection Algorithm Based On The Feature Extraction Algorithm Of LSK

The target detection technology in image is an important research section in the field of computer vision technology.In recent years,with the rapid development of Internet and computer,target detection technology has been widely applied in medical exploration,video surveillance,military target search,inspection robot and other popular areas.A complete set of target detection algorithm should have a high degree of autonomy,high efficiency,high stability,high accuracy and so on,which can directly detect and locate the target from the collected images.With the maturity of the target detection technology,more and more inspection robot is equipped with the target detection module,and the instrument detection is one of the important research contents of the target detection.The complexity of the scene environment,such as the influence of natural light,the interference of a large number of similar objects and some other factors,has brought great challenges to the instrument detection technology.The detection and localization of instruments includes three main steps:image feature extraction,feature dimension reduction and feature matching.Based on the analysis and research of the traditional detection methods of the visual target,combining the LSK feature extraction algorithm and different dimensionality reduction algorithms,this paper proposes a method of instrument detection and localization under the complex scene.And experiments are carried out to verify the validity of the method.The main work of this paper includes:(1)The basic theory of image target detection and localization is analyzed and studied,and the research status and methods of image target detection and localization are described.(2)This paper briefly introduces the method of image target feature representation,with emphasis on the histogram of oriented gradient(HOG),the local binary pattern(LBP)and the local steering kernel(LSK).(3)The dimensionality reduction method of image feature is studied,and the effectiveness of different dimensionality reduction algorithms for LSK feature is verified by using face data as samples.(4)This paper presents a method for detecting and locating instrument in complex scene,in which,the local steering kernel function(LSK)is used to extract the feature of the image,and the neighborhood preserving embedding(NPE)is used to reduce the dimension of the feature vectors.The experimental results show that the proposed method has good performance.