Study On The Method Of Specific Target Recognition In Complex Environment

Speci fic target recogn ition t echnology is one o f the impo rtant research direction of co mp uter vision and static ima ge processing technology.Target recog nition is used fo r mil itary and defense fields. In recent years, wi th the popularization of vis uali zed monit oring sys tem, the specific target recognition t echnology is gradually appli ed i n the monitorin g syst em based on image proces sing.At pres ent, more and more s pecific t arget object p lays a more import ant role in human life, such as cellphone, l aptop, USB flash drive and o ther types o f mobile i nformation s torage device. These target object devices continues to be smaller and more portab le. But meanwhile, these objects also beco mes more easily l ost, easily s tol e n, hard t o control and so on.This paper b ased on specific target recogniti on and control in a complex environment as the d irectio n, the s tudy analyzed the current situation of the development of st ati c imag e t a rget recognition and some defects and shortcomi ngs in the current applicat ion. Then focusing on t he anal ysis of SIFT feature descriptor based on feature detectio n, doin g anal yses and simulat ion ex periment using SIFT algorithm o n s pecific target, studying on SIFT, SURF algorithms of the p resent applicat ion in advantages and disadvant ages, doi ng some optimi zati on of the algori thm.Keeping mo re effective features, redundancy remov al, reduce the amount of computat ion in o rder to improve the matching speed target recogn ition.After that, thi s paper ad opts the mode of mi xed Gau ssian backgrou nd mod eling, p roposed a n ew met hod o f mo vin g o bject detection in co mp l ex environmen t of visual mo nitoring sys t e m, and cooperate with the targetrecog nition algori thm t hrough calculating an d processing of certain to achieve t he goal of contro lling specific target object.Finally, throu gh simulation tests accordi ng to real scene, it verify t he feasibility and efficiency of the improved algori thm. Co mpared with t he existin g algorithms, this method for specific target recognition under complex environment has a high er rate of fault tolerance, and get a better recogniti on efficien cy.