Visual attention and mean shift-based target detection and tracking method

Abstract

The invention discloses a visual attention and mean shift-based target detection and tracking method. The method comprises the following the steps: firstly, extracting the salient region of a first frame image in an image sequence by using a visual attention method, and removing interferences of background factors to obtain a moving target; then, changing fixed bandwidth of a kernel function in a traditional mean shift method into dynamically changed bandwidth, and tracking the detected moving target by using an improved mean shift method. Shown by an experimental result, the visual attention and mean shift-based target detection and tracking method disclosed by the invention is suitable for infrared and visible image sequences, and better in tracking effect. Moreover, the positional information of the moving target also can be provided by the visual attention and mean shift-based target detection and tracking method disclosed by the invention, and thus, possibility is provided for accurate positioning of the target. The visual attention and mean shift-based target detection and tracking method has a broad application prospect in the military and civil field of night vision investigation, security and protection monitoring, and the like.

Description

technical field [0001] The invention relates to a method in the technical field of target tracking processing, in particular to a target detection and tracking method based on visual attention and mean shift, which can be used for infrared and visible light image sequences. Background technique [0002] Target detection and tracking technology is to detect the region of interest (usually a moving target) in a sequence of video images, and obtain some relevant information about it in real time, such as position, size, speed, acceleration, etc. [0003] At present, most mature target detection and tracking methods are based on pixel level, which can be roughly divided into the following two categories: [0004] (1) Detect first and then track, without relying on prior knowledge, directly detect the target from the video image sequence, extract the target of interest, and finally realize the tracking of the moving target. Such as threshold segmentation, filters based on morpho...

AI Technical Summary

Problems solved by technology

[0006] The detection-then-tracking method has a relatively simple structure and is easy to implement. It has better tracking performance when the signal-to-noise ratio is relatively high. However, when the signal-to-noise ratio is low, lowering the threshold will result in a higher false alarm rate. Moreover, it will increase the burden of the follow-up track correlation method, which will greatly increase the amount of calculation; while the tracking first detection method is based on the detection and tracking technology of the target's motion characteristics, and uses the motion characteristics to describe the target. Relatively speaking, the amount of calculation is relatively large. Many methods of this type cannot guarantee the real-time performance of the tracking effect, and the reliability of the method needs to be improved

[0007] When using the traditional visual attention method to detect moving targets, the amount of calculation is large, and the processing time for each frame of image is long, it is difficult to meet the real-time requirements, and the interference of background factors cannot be ruled out; when using the traditional mean shift method to track the target , using the color histogram of the target as the search feature, and gradually converging on the real position of the target through continuous iteration, so as to achieve the purpose of tracking. This is a semi-automatic target tracking method, which needs to select the tracking target in the initial frame, and Build a Probabilistic Model

In addition, when the target moves fast or the target is occluded, the traditional mean shift method usually cannot search for the target position

For the combination of the two image processing methods of visual attention and mean shift, domestic scholars have also done work in this area. First, the visual attention method is used to process the image to obtain the salient area, and then the mean shift method is used in each salient area. Iteratively get the matching area, so as to achieve the purpose of target tracking. This method has a better result for the target moving too fast, but each frame of image must first use the visual attention method to obtain the salient area, and then use the mean shift method to track the target. The amount of calculation is very large, and the real-time requirements that are very important in the field of target tracking cannot be met

Images