Research On Efficient Object Detection Algorithm For High Speed Sorting

Target recognition and localization is one of the most important parts of machine vision research.As an effective means of target recognition and location,template matching is currently widely used in industrial identification and location of objects,tracking and integrity testing and other occasions.With the rapid development of industrial automation,the requirements for machine vision are also increasing.Therefore,it is important to develop a template matching algorithm that meets industrial applications.This paper presents a template matching algorithm which based on geometric features.Make edge point gradient direction as the matching information of the algorithm,in order to ensure high-speed matching,using the image pyramid from coarse to fine matching strategy and integral images in a timely manner to exit non-target location.At the same time,so as to obtain high precision,the edge fitting is used to extract sub-pixel-level edge information.Before the threshold judge start the feature point sets are discretely sorted,and good robustness can be guaranteed even when the target exists partially occluded and hidden.The main research contents are as follows:First,the two processes of coarse and fine matching are introduced,and the design of fast template matching algorithm is introduced in this paper.The similarity measurement method is determined,and the normalized gradient value method is used to avoid the influence of illumination.At the same time,using integral image and image Pyramid to speed up.The principle of surface fitting and the acquirement of sub-pixel precision edge are also introduced.Secondly,this paper introduces several key algorithms,the template to create the feature points of surface fitting to obtain sub-pixel edge and mean filtering method is chosed after analysising the advantages and disadvantages of the method of constructing image Pyramid.Besides,this algorithm chose a different angle the pace of rotation of the template set of feature points in the different layers of Pyramid,and draw the flow chart of the algorithm of the matching process.At the end of the visual system hardware platform is introduced and the stability and reliability of algorithm and the matching speed and accuracy are tested,to understand the actual performance of the algorithm,this paper carries on the summary of research results pointed out the deficiency and puts forward the improvement direction.