An object tracking method and system combining channel selection and position optimization

Abstract

The present invention discloses an object tracking method and system combining channel selection and position optimization, which relates to the field of object tracking. The method includes: obtaining a video sequence to be tracked; using a CNN network model to perform feature extraction on the video sequence to be tracked to obtain convolutional features; the convolutional features include a third convolutional feature, a fourth convolutional feature, and a fifth convolutional feature; sequentially performing gradient calculation, channel selection, and cross-correlation operations on the convolutional features to obtain a first target response map; determining an initial target localization position and a peak sidelobe ratio according to the first target response map; performing position optimization according to the initial target localization position and the peak sidelobe ratio to obtain a final target localization position. The present invention takes into account both tracking accuracy and speed, and solves the problem of tracking drift in scenarios where the target undergoes large deformations or has low resolution, etc.

Images