Target tracking method based on supporting online clustering in detection

Abstract

The invention provides a target tracking method based on supporting online clustering in detection, and belongs to the technical field of a computer graphical image. The target tracking method comprises the following steps of: simultaneously storing characteristic vectors of a target and a background sample in a blade node of a random fern detector; exploring distribution characteristics through the online clustering; and carrying out type probability density estimating by taking the distribution characteristics as sampling data points of a core function. A target object to be tracked is selected and determined from an original image, and is added into an online model consisting of target image blocks; and a target selecting process can be automatically extracted through using a motion target detecting method. In a real-time treatment condition, a video image collected by a camera and stored in a storage area is extracted as an input image to be tracked. The target image block is extracted as a positive sample, and a background image selecting block is used as a negative sample, so that an online training set is generated and input to the detector. A plurality of target types exist according to a condition of a plurality of targets, and each target type corresponds to one target. The target tracking method is mainly used for various fields of target tracking.

Description

technical field [0001] The invention belongs to the technical field of computer graphics and images, and in particular relates to computer vision, pattern recognition and machine learning technologies. Background technique [0002] Since object tracking is a fundamental problem in many computer vision applications, such as intelligent video analysis, video surveillance, human-computer interaction, etc., researchers have done a lot of work on it, but so far it has been difficult to achieve long-term Visual tracking is still an extremely challenging task. With the development of stochastic computer vision technology, object tracking methods based on machine learning, especially online learning, have received more and more attention. This is due to: (i) In general, before object tracking begins, we can only obtain very limited prior information for building the necessary knowledge of the tracking system, such as object appearance, background information, and motion model. A l...

AI Technical Summary

Problems solved by technology

However, like random forest, random fern needs a large amount of data as training samples and tests before it can give full play to its classification ability, so random fern shows insufficient initial classification ability in object tracking and it is difficult to recover after tracking failure; On the other hand, the evaluation process of random fern on test samples simply depends on the number and type of training samples in the corresponding leaf nodes, without considering the spatial distribution of feature vectors, which largely affects the random fern's Classification or detection accuracy

Images