Automatic spherical camera tracing method

Abstract

The invention provides an automatic spherical camera tracing method, comprising the steps of, acquiring an image; acquiring a motion vector of an image; determining a motion state of the spherical camera; performing binaryzation of a scene image; segmenting connected objects in the motion vector; analyzing and tracing the segmented connected objects; and tracing a target object. The method of the invention uses the motion vector to assist in segmenting foreground objects. With respect to a macroblock of N*N pixels, pixels which need to be treated are 1 / N*N of pixels of an original image. And for application of automatic tracing, only one object needs to be traced at a time, and the method can satisfy the requirement already, so hardware load can be reduced largely. The method is suitable for an embedded platform with poor hardware efficiency, and the spherical camera can run more fluently without an extra signal processing integrated circuit. Even if the target object suddenly accelerates, decelerates, is covered by barriers or overlaps with another object, a good tracing effect can be guaranteed.

Description

technical field [0001] The present invention relates to a method for automatic tracking of a dome camera, especially a method of using motion vectors to assist in cutting foreground objects, thereby reducing the burden on hardware and making the dome camera run more smoothly. And has an excellent tracking effect, and is suitable for monitors, camcorders, dome cameras or similar devices. Background technique [0002] At present, the surveillance system is quite diverse. Many indoor or outdoor video surveillance systems use surround cameras under the limitation that the existing fixed cameras cannot cover all surveillance areas. Although the surveillance range of the surround cameras is larger than that of traditional fixed cameras It is much larger, but as the monitoring area becomes larger, the difficulty in image quality control also increases. The difficulty comes from the fact that the lens of the surround camera cannot be zoomed in or out, and moving objects in the moni...

AI Technical Summary

Problems solved by technology

[0004] Among the existing target object tracking technologies, the most traditional one with the longest development history is the automatic object feature tracking technique, which mainly uses feature extraction and feature comparison. The method of matching) extracts image features from the target object, and then locks and tracks the target object according to the image features. However, when the target object moves, the background of the monitoring area usually changes, and when the target object rotates, the image features also In addition, the continuous scanning action of the PTZ camera, all the above factors will cause the difficulty of the actual operation of the feature comparison method

[0005] Another existing target object tracking technology uses two cameras, uses the epipolar rule, and calculates the projection fundamental matrix (projection fundamental matrix) according to the feature points of the corresponding imaging surface. ), to lock and track the target object, but when one of the cameras performs scanning and rotates, the corresponding imaging surface also changes, and the corresponding relationship between the two cameras has changed. It is not easy to correspond to the feature points of

[0006] Furthermore, from the perspective of the development history of tracking algorithms, many effective methods have been proposed, and the CamShift algorithm is based on the color probability distribution to track, by calculating the color probability distribution and continuously iteratively converging Then, to find the center and size of the target, and then find the target to be tracked, but its disadvantage is that if you encounter an object of similar color, such as a person who is tracking, the color is red, and the color of the chair is just red , when a person walks near the chair, the CamShift algorithm may jump to a chair with a similar color, causing tracking errors, and because the CamShift algorithm needs to calculate the color probability distribution, it requires a lot of data processing and calculation, so additional Signal processing integrated circuit (DSP, Digital Signal Processing digital signal processing) to assist or use a high-performance computer to assist in the calculation

Images