Object tracking method and system based on depth characteristic flow, terminal and medium

Abstract

The invention provides an object tracking method and system based on a depth characteristic flow, a terminal and a medium; the method comprises the following steps: using a visual sensor to capture acontinuous video frame sequence, and obtaining the current frame image from the video frame sequence; detecting a mobile object, determining at least one key frame containing the mobile object, building a mapping relation between the current frame image and corresponding characteristics in the key frame image, and obtaining an optical flow estimated vector graph; respectively filtering static objects in the key frame image so as to obtain a motion characteristic extraction network related to the mobile object, and extracting a characteristic graph; propagating the key frame mobile object characteristics onto the current frame. The method employs the depth characteristic flow to process images, is fast in speed, cannot affect video segmentation and task identification, and can use an end-to-end mode to map, thus improving precision.

Description

technical field [0001] The present invention relates to the technical field of vehicle electronics, in particular to a method, system, terminal and medium for object tracking based on depth feature flow. Background technique [0002] ADAS is the advanced driver assistance system, also known as the active safety system, mainly including the electronic stability system ESC, the adaptive cruise system ACC, the lane departure warning system LDW, the lane keeping system LKA, the forward collision warning system FCW, the door opening warning DOW, the automatic Emergency braking system AEB, traffic sign recognition TSR, blind spot detection BSD, night vision system NV, automatic parking system APS, etc. [0003] The ADAS active safety system must not only recognize static objects, but also recognize dynamic objects. At present, the deep convolutional neural network has achieved great success in image recognition tasks. However, it is a very complicated task to convert the image re...

AI Technical Summary

Problems solved by technology

Second, the optical flow estimation uses the bilinear interpolation method to propagate the feature map of the key frame to the corresponding position of the current frame. In this way, after a key frame and the feature map obtained by two frames are given, the key The feature map of the frame is propagated, that is, the feature map of each pixel is copied to the position corresponding to the current frame, which can solve the unsatisfactory situation of many frames in the video such as occlusion and motion blur.

Images