Motion object detection method combining LK optical flow method and three-frame difference method

Abstract

The invention discloses a motion object detection method combining an LK optical flow method and a three-frame difference method. The motion object detection method comprises steps of 1) determining an area where a motion object could exist through a pyramid type LK optical flow method, 2) using the area determined in the step 1, wherein a threshold value of a three-frame differencing is a lower value inside the area and the threshold value of the three-frame differencing is a higher value outside the area, and determining the difference of the threshold values is a difference value of the higher threshold value and the lower threshold value, 3) choosing a start angular point of an optical flow vector which is calculated by the LK optical flow in a low threshold value area and taking the start angular point and an area obtained by detecting through the three-frame difference method after eliminating part of error angular point information as motion objects, 4) performing processing through a connection analysis method in a low threshold value area to calculate an area of each connection area, setting a smaller threshold value on the area, if the area is smaller than the threshold value, filtering the connection area.

Description

technical field [0001] The invention specifically relates to a moving target detection method combining an LK optical flow method and a three-frame difference method. Background technique [0002] Moving target detection is to separate the moving target in the video sequence from the background image, so that the foreground of the target can be obtained, that is, the exact moving target. It is a key technology in the field of computer vision research and belongs to the bottom layer of the intelligent video surveillance and target tracking system. The quality of the detection results is closely related to the effect of subsequent target tracking and behavior understanding. [0003] At present, the commonly used moving target detection algorithms include background subtraction method, optical flow method and frame difference method. The optical flow method is to separate the moving target from the background by estimating the motion field generated during the motion of the ob...

AI Technical Summary

Problems solved by technology

The biggest advantage of this algorithm is that it is fast, the background does not need to be accumulated, and it is not sensitive to illumination changes. The algorithm complexity is low, but it is very sensitive to environmental noise. A lower threshold value will increase the noise, while a higher threshold value will increase the noise. Make the moving target incomplete, so it is difficult to suppress the noise in the image and retain the effective and complete moving target