A Moving Object Detection Method Based on Optical Flow and Frame Matching

Abstract

The invention relates to a movement target detecting method based on light streams and inter-frame matching. Two frames of images are preprocessed and roughly calibrated, and an overall environment deviation component is calculated; the light stream field of the two frames of images is calculated; the overall environment deviation component is subtracted, and the light stream field movement amplitude is obtained; the light stream field movement amplitude is converted into a threshold, and the movement target area and a non-movement area without the overall environment deviation component are obtained; the two frames of images are precisely matched, and a perspective transformation matrix from the previous frame of image to the subsequent frame of image is calculated; the previous frame ofimage is converted to a coordinate system of the subsequent frame of image through the perspective transformation matrix; differentiation is conducted on the two frames of obtained images, threshold division is conducted on the differentiation result, denoising is conducted through a morphological algorithm, and a plurality of continuous movement target areas are obtained; the continuous movementtarget areas are clustered, and central coordinates of the areas are obtained; the continuous movement target areas with the distances between the central coordinates being smaller than a set threshold are combined to be output, and a detected movement target is obtained.

Description

Technical field [0001] The invention relates to a moving target detection method based on optical flow and frame matching, which is suitable for the detection field of moving targets in complex scenes of high-definition images. Background technique [0002] Target detection technology is an important technology in the field of digital image processing and artificial intelligence. It is dedicated to segmenting candidate targets from a complex background for further tracking and recognition. Target detection is to locate the position and size of the target of interest in the background. According to the target motion situation, it can be divided into two categories: static target detection and moving target detection. [0003] The common methods of moving target detection include: inter-frame difference, background difference and motion segmentation. Inter-frame difference uses the change between frames to realize detection. Common methods include two-frame or three-frame differenc...

AI Technical Summary

Problems solved by technology

[0009] The technical problem solved by the present invention is to provide a method based on optical flow and inter-frame matching for the detection of moving objects in complex scenes, and to improve the precision and accuracy of object detection by combining the motion segmentation method and the method of first matching and then difference

Images