Three-dimensional Hough transform-based track-before-detect method and radar target detection system

Abstract

The invention belongs to the technical field of target detection and tracking or tracking before target detection, and discloses a three-dimensional Hough transform-based track-before detect method and a radar target detection system. According to the method, trace points cohered by a video and obtained in a period of time are input, and each trace point comprises two-dimensional space informationand one-dimensional time information; two trace points are randomly selected, and through the two trace points, a three-dimensional space straight line can be uniquely determined; three parameters are used to define a space straight line, and when voting is carried out in Hough transform, a voting space is three-dimensional; all the trace points are combined in pairs to obtain three-dimensional straight lines, the straight lines are mapped to the voting space, and a straight line parameter which has most votes and is greater than a detection threshold is a detected target track; trace pointsin the navigation track are deleted and the operation is repeated to obtain all the target tracks in sequence. By means of the Hough transform, the method is capable of correctly detecting multiple targets under the condition of high false alarm so as to obtain a track set.

Description

technical field [0001] The invention belongs to the technical field of target detection and tracking or target tracking before detection, and in particular relates to a three-dimensional Hough transform-based tracking-before-detection method and a radar target detection system. Background technique [0002] At present, the commonly used existing technologies in the industry are as follows: target detection and tracking play a very important role in video tracking, radar data processing and other aspects. Among them, the use of multi-frame data for joint detection can effectively deal with the high false alarm rate, and this method is also called the tracking before detection algorithm. In recent years, this algorithm has rapidly become the focus and hotspot in target detection and tracking due to its good detection performance. The existing tracking-before-detection algorithms include three categories, namely tracking-before-detection algorithm based on particle filter, tra...

AI Technical Summary

Problems solved by technology

[0004] (1) The tracking-before-detection algorithm based on particle filter has insufficient detection ability for multiple mutually adjacent targets, and the detection and tracking effect is not good in dense target areas, resulting in target missed detection and tracking divergence

[0005] (2) The tracking-before-detection algorithm based on dynamic programming is not suitable for extended targets, that is, when a target generates multiple traces in each frame, using this method will generate a large number of repeated targets; it is only suitable for uniform periodic situations such as For a radar that rotates at a uniform speed, for a radar in a non-uniform period such as a sector scan mode, the video information is not uniformly processed in the direction of the scan line, and the track detection and tracking cannot be performed according to the original algorithm

[0006] (3) Most tracking-before-detection algorithms based on Hough transform use three two-dimensional Hough transforms to simulate three-dimensional Hough transform, which may easily cause problems such as missed detection and false track.

The five-dimensional parameter field has been tested, and due to the existence of measurement errors, the votes of the same track are too scattered, resulting in poor detection results

At the same time, there are too many parameter cells in the five-dimensional parameter domain, resulting in a high amount of calculation

Images