Multi-feature Target Adaptive Tracking Algorithm Based On The Confidence Of Feature

As an important part in computer vision, target recognition andtracking has been widely used in various fields, such as intelligent robots,intelligent transportation, navigation, surveillance and military fields. It isthe main content of automatic control, computer vision and patternrecognition. With the technology of LabVIEW, this paper achieved thetracking algorithm based on color feature, shape feature and motion feature.In addition, it also proposed the confidence of two types of trackingfeatures, demonstrate their rationality, and give an adaptive multi-featuretarget tracking algorithm and experimental verified.The main work of this article as follows:(1) Build a target tracking system based on the color feature of thetarget with the technology of LabVIEW, and achieved the desired results.(2)Build a target tracking system based on the shape feature of thetarget with the technology of LabVIEW, and achieved the desired results.(3) Put forward a kind of evaluation method about the confidence ofcolor feature and shape feature, and demonstrates its reasonableness.(4) Aimed at the tracking algorithm based on multi-feature fusionreal-time performance is poor, This article proposed a tracking methodbased on the confidence of features, which can switch according to themost reliable characteristics of the tracking feature. This method hadcharacteristics of low computation complexity and excellent real-timeability.