出版社：Academy & Industry Research Collaboration Center (AIRCC)
摘要：This paper proposes a robust tracking method which concatenates appearance and geometricalfeatures to re-identify human in non-overlapping views. A uniformly-partitioning method isproposed to extract local HSV(Hue, Saturation, Value) color features in upper and lowerportion of clothing. Then adaptive principal view selecting algorithm is presented to locateprincipal view which contains maximum appearance feature dimensions captured from differentvisual angles. For each appearance feature dimension in principal view, all its inner frames getinvolved in training a support vector machine (SVM). In matching process, human candidatefiltering is first operated with an integrated geometrical feature which connects height estimatewith gait feature. The appearance features of the remaining human candidates are later testedby SVMs to determine the object’s existence in new cameras. Experimental results show thefeasibility and effectiveness of this proposal and demonstrate the real-time in appearancefeature extraction and robustness to illumination and visual angle change.
