Online adaptive adjustment tracking method for target image regions

Abstract

The invention relates to an online adaptive adjustment tracking method for target image regions. The method includes the steps of acquiring Haar features from different estimated positions in a newly input video frame, calculating likelihood of the image regions at the positions by a Boosting structured classifier, and using one image region with highest likelihood as a target image region in the current frame. The softer is updated online by the aid of a weak classifier pool and an auxiliary classifier pool, and adaptivity of the classifier to appearance change of a target is further improved. Probabilities of occurrence of background and target image samples are asymmetrical, distribution weights of the samples are adjusted according to classification errors of the weak classifiers, and accordingly the classifiers are highly sensitive to occurrence of targets in the video frames. Therefore, the targets in video can be tracked more stably by the method.

Description

technical field [0001] The invention relates to an online self-adaptive adjustment target image area tracking method. Background technique [0002] Target tracking can be considered as a binary classifier problem between the background and the target object in the image. The classifier constructed on the basis of the Boosting algorithm can obtain high discriminative features about the specific target object in the image. Some scholars have proposed the application of this method In the field of target tracking, Grabner uses the Boosting method for feature selection, thereby constructing an appearance model that can be updated incrementally. Advian selects a group of classifiers from the pool of weak classifiers through the Adaboost method to detect the most likely image area of ​​the target object in the video image. The above method is to minimize the classification error under the assumption that the occurrence probability of the background sample and the target sampl...

AI Technical Summary

Problems solved by technology

The above method is to minimize the classification error under the assumption that the occurrence probability of the background sample and the target sample are equal. When the probability of the background sample is much greater than the target object, especially when the target object is occluded, the classifier constructed by these methods is easier. Discriminate the target as the background, which directly leads to the failure of tracking

Images