关键词: 感知哈希算法, 目标跟踪, 遮挡, 目标丢失, MeanShift

Abstract:

Traditional target tracking algorithms suffer high computational complexity, occlusion, shape change, and loss of target motion. The image perceptual hash algorithm is used in target tracking and improved for the special target of pedestrians. First pedestrians are divided according to their characteristics into several regions that are assigned different weights. The perceptual hash value of the tracking target is calculated, and the diamond search algorithm is used to calculate the area of the hash value of the unknown. Finally the suitable area is chosen to be tested as the target region. This algorithm has better continuity of target occlusion and tracking than the MeanShift algorithm.

Key words: perceptual hash, target tracking, occlusion, target loss, MeanShift