A method and device for video complex scene region segmentation based on decision-making layer fusion

Abstract

The present invention provides a video complex scene region segmentation method and device based on decision-making layer fusion. The method and device comprehensively utilize the hue feature value, texture feature value and dynamic feature value of pixels in different image frames of the video, and The eigenvalue dimension and the eigenvalue range are constrained, and the statistical time span of the dynamicity of the pixel is limited to avoid the occurrence of a situation where the statistical time span is too large to reduce the accuracy of dynamic statistics. The fusion strategy of the decision-making layer rather than the feature layer processes the scene region segmentation processing results about different feature values. In addition, a parallel computing mode is introduced to obtain the hue feature value, composite texture feature value and recent dynamic feature value at the same time, which can significantly improve The efficiency of its overall operation, thereby improving the effectiveness, correctness and operation efficiency of video complex scene area segmentation.

Description

technical field [0001] The invention relates to the technical field of video processing, in particular to a method and device for segmenting video complex scene regions based on decision-making layer fusion. Background technique [0002] With the vigorous development of technologies such as visual robots and unmanned driving, the importance of vision-based scene analysis technology has become increasingly prominent. One of the underlying technologies in vision-based scene analysis is field of view segmentation. The result of field of view segmentation will directly determine the reliability and accuracy of upper-layer scene type recognition and object recognition in the scene. [0003] In the related technology, the color feature vector of the RGB color space of each pixel and the color feature vector of the HSV color space are connected in series to generate the color feature vector of the dual color space of each pixel, and then the dual color feature vector of each pixel ...

AI Technical Summary

Problems solved by technology

[0004] However, the above-mentioned technology has the following defects. First, the above-mentioned technology does not limit the value range of color features. In complex content scenes with rich color types, the number of categories of sub-region features generated by feature layer fusion is huge, which makes it unsuitable for color features. Scenario analysis of complex content; second, the sub-region feature dimension and value range generated by the series of the above technologies are too long, resulting in a huge amount of subsequent clustering calculations, too long time-consuming region segmentation and poor real-time performance, and it is easy to cause Information redundancy and increase the fusion feature dimension of the feature layer; third, the pixel dynamics calculated by the above technology are for the average value of the entire video time span, which makes it not suitable for processing long videos with a large time span

Images