A method and system for target tracking based on dynamic measurement matrix

A dynamic measurement and target tracking technology, applied in the field of computer vision, can solve problems such as feature mode solidification and tracking drift, and achieve strong adaptability and good robustness

CN105096345BActive Publication Date: 2017-11-07UNIV OF ELECTRONICS SCI & TECH OF CHINA
1 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Publication Date
2017-11-07

Smart Images

  • Figure 1
    Figure 1
  • Figure 2
    Figure 2
  • Figure 3
    Figure 3
Patent Text Reader

Abstract

The invention discloses a target tracking method and system based on a dynamic measurement matrix, which includes the following steps: S1: compress the high-dimensional features of the sample into low-dimensional features, and initialize the dynamic measurement matrix; S2: collect multiple A set of positive samples and a set of negative samples are used to update the classifier; S3: determine the position of the target in the current frame; S4: update the dynamic measurement matrix, and return to step S2 until the tracking is completed. The present invention uses the dynamic measurement matrix to extract the compressed features of the target in the tracking process, that is, in the tracking process, the feature of the tracked target and the characteristics of the naive Bayesian classifier are used to update the measurement matrix, and the fixed form of the measurement matrix is ​​improved. This tracking method has strong adaptability. Experimental results prove that the tracking method has good robustness.
Need to check novelty before this filing date? Find Prior Art

Description

Technical field

[0001] The invention relates to the field of computer vision, in particular to a target tracking method and system based on a dynamic measurement matrix. Background technique

[0002] Target tracking based on video or image sequence is one of the hot issues of computer vision, and it is very closely related to target detection. Tracking-by-detection methods also benefit from the efficiency and accuracy of detection problems. Promote. There are many difficulties to deal with the target tracking problem, such as lighting changes, target shape or posture changes, and complex scenes, which have a great impact on the tracking effect.

[0003] Tracking problems can be roughly divided into two categories. One type is tracking based on generative models. This type of method learns the appearance model of the target. When searching for the target position, it looks for the area that is closest to the model or has the smallest reconstruction error; the other type is trackin...

Examples

Embodiment Construction

[0055] The technical solution of the present invention will be described in further detail below in conjunction with the accompanying drawings: figure 1 As shown, a target tracking method based on a dynamic measurement matrix includes the following steps:

[0056] S1: Compress the high-dimensional features of the sample into low-dimensional features, and initialize the dynamic measurement matrix;

[0057] S2: Collect multiple positive sample sets and negative sample sets around the target location, and perform classifier update learning;

[0058] S3: Determine the position of the target in the current frame;

[0059] S4: Update the dynamic measurement matrix, and return to step S2 until the tracking is completed.

[0060] Step S1 includes the following sub-steps:

[0061] S11: Put the sample High-dimensional features Compressed into low-dimensional features which is:

[0062] v=R(t)x;

[0063] In the formula, where m = (wh) 2 , N Is the dynamic measurement matrix;

[0064] S12: Initiali...