Moving target extraction method based on difference and semantic information fusion

Abstract

The invention discloses a moving target extraction method based on difference and semantic information fusion. The method mainly comprises the following steps: (1) collecting N frames of image sequences from monitoring equipment; (2) calculating difference information between image frames by using an inter-frame difference method according to two frames of images with an interval of N; (3) extracting semantic information in the image by using the trained instance segmentation model based on the convolutional neural network, wherein the semantic information comprises a target category and a pixel mask; and (4) combining the differential information and the semantic information through a fusion algorithm, and extracting a moving target in the image. According to the method, strong semantic information is introduced through an instance segmentation model based on a convolutional neural network, and a moving target in an image can be well extracted in combination with difference information obtained by an inter-frame difference method. The method is simple to implement and has good robustness.

Description

technical field [0001] The invention relates to the fields of digital image processing and computer vision, and specifically designs a moving target extraction method based on difference and semantic information fusion. Background technique [0002] The analysis of moving targets has always been one of the important research contents in the field of computer vision, and has a wide range of applications in production and life. The most common application scenario is the analysis of surveillance video. Usually, static targets are not paid much attention to, but moving targets need to be analyzed, because moving targets are likely to cause important impacts on production and living activities in the current monitoring scene. influences. The current analysis of moving targets is mainly based on the background subtraction method and the inter-frame difference method (Jodoin, Pierre-Marc. Comparative study of background subtraction algorithms [J]. Journal of Electronic Imaging, 2...

AI Technical Summary

Problems solved by technology

At present, most convolutional neural networks are predicted for a single image, so it is difficult to obtain motion information in the image (Braham M, Sébastien Piérard, Droogenbroeck M V. Semantic Background Subtraction[C] / / IEEE International Conference on Image Processing. IEEE,2018.

Images