Moving target detection method based on space-time clustering

Abstract

The invention aims to provide a moving target detection method based on space-time clustering, which can give consideration to calculation speed and detection stability, quickly distinguish a dynamic target and a static target from an image sequence, and track the dynamic target to determine the position of the target in an image. The moving target detection method comprises the following steps: (10) extracting image feature points: carrying out corner detection on a target in an initial frame, and extracting the image feature points; (20) feature point track association: performing tracking detection on the extracted image feature points, and performing track association; (30) track displacement clustering: according to different displacements of background motion and target motion in image coordinates, cutting out motion feature points and background feature points through clustering; and (40) motion feature point spatial clustering: according to spatial differences of different targets in the image, combining feature point tracks on the same target, distinguishing feature point tracks on different targets, and recording the track state of each motion target at the same time.

Description

technical field [0001] The invention belongs to the technical field of moving target detection, and proposes a target extraction algorithm for quickly distinguishing dynamic targets and static targets. Background technique [0002] Moving target detection refers to the ability to identify image changes in a specified area, detect the existence of moving targets and avoid interference caused by light changes. Motion detection such as MPEG1 is to compare and analyze the key frames generated after encoding. The principle is to divide the image sequence into key frames (I), predicted frames (P frames) and double interpolated inward frames (B frames). Process comparisons are performed on the encoded data. Common methods of data processing include: [0003] (1) Background subtraction: Background subtraction is one of the most commonly used methods, which uses the difference between the current image and the background image to detect motion regions. It can generally provide th...

AI Technical Summary

Problems solved by technology

[0007] In short, the problems existing in the existing technology are: poor anti-noise performance, inability to distinguish different targets, and high algorithm complexity

Images