Infrared moving target detection method based on structural similarity and saliency analysis

Abstract

An infrared moving target detection method based on structural similarity and saliency analysis in the field of image processing, using GBVS to analyze the saliency of the source image to obtain the saliency area; Gentle area, and the Gaussian mixture model is optimized by different learning rates; finally, the closed area is detected by the optimized Gaussian mixture model, and the overlapping area between the closed area and the salient area is the final moving target of the current frame. The invention solves the problem of blurred edges of the mixed Gaussian algorithm, and has good adaptability and detection effect.

Description

technical field [0001] The invention relates to a technology in the field of image processing, in particular to an infrared moving target detection method based on structural similarity and saliency analysis. Background technique [0002] Real-time detection of moving targets is the premise of follow-up target tracking and recognition, and its effect directly affects the robustness and accuracy of follow-up work. Moving object detection refers to the extraction of foreground objects that are relatively moving with the background in the video sequence, and prepares for more advanced motion analysis such as subsequent object tracking. It is an important research direction in the field of computer vision. It is used in intelligent monitoring systems, human It has been widely used in computer-computer interaction systems and other fields. [0003] Moving object detection detects moving objects by analyzing the image sequence captured by the image sensor, laying a good foundatio...

AI Technical Summary

Problems solved by technology

[0006] The present invention aims at the deficiency that the traditional Gaussian mixture model cannot detect the change of the background in time, and the infrared target obtained through its detection contains false contours and is difficult to be accurately identified, and proposes an infrared moving target detection method based on structural similarity and saliency analysis and The system uses the watershed algorithm combined with the spatial information representing the discrete pixels of the target area to obtain a closed area, and then eliminates the ghost through the segmentation algorithm based on PCNN (Pulse Coupled Neural Network, Pulse Coupled Neural Network), and finally detects the complete infrared moving target. In this way, the foreground target can be better extracted, with better effect and stronger robustness.

Images