Moving Target Detection And Tracking Algorithm Research Based On OpenCV

As a new field of research,target detection and tracking has acted a more andmore important role on science and engineering since21st century and it has a verywide range of applications in our lives. It is widely used in video surveillance,intelligent human-computer interaction,medical diagnostics, virtual reality, andmilitary fields. In this thesis, the main research content includes: how to use thehardware platform of personal computer and related algorithm to get related datafrom the moving objects in a video image sequence to effectively realize the detectionand tracking of moving targets.This paper mainly studies the detection and tracking algorithms of moving targetand does the following works:(1)This paper based on VS2010software platform for algorithm design bycalling the OpenCV library functions. First of all, properly configured OpenCVplatform, and then introduced the OpenCV libraries, finally completed theconstruction work of the experimental environment.(2)For gaussian noise and salt-pepper noise,preprocessed the image, introducedseveral filtering methods for removing noise, deeply analyzed of the mean filteringalgorithm and median filtering algorithm; and combined the image corrosiontechnology and image expansion technique to study the image restoration technology.(3)In the simple static background, deeply analyzed and discussed the framedifference method,background subtraction method and optical flow method thesethree kinds of moving target detection algorithms. Considered the advantages anddisadvantages of each algorithm, combined with the background subtractionalgorithm and the frame difference algorithm respective advantages, used thedetection method based on gaussian mixture model. Through experiments it can beconcluded that the algorithm has adaptively got the background image and excludedthe interference of moving targets, successfully detected the moving targets, andachieved good detection effect.(4)Deeply researched the CamShift algorithm and Kalman filtering algorithm,forthe limitations of CamShift algorithm,combined the CamShift algorithm and Kalman filtering algorithm to realize tracking moving objects. The experiments showed thatthis method can effectively improve the tracking effect and accuracy.(5)In the moving target detection and tracking experiments,. after running andverification,the following conclusions were drawn: for the average execution time ofeach frame in the sequence images,the accuracy、real-time、and high efficientrequirements can be satisfied when the time is30ms.