Occlusion target recognition method of sar image fusion based on sparse representation of segmented images

Abstract

The invention discloses a SAR image fusion occlusion target recognition method based on sparse representation of segmented images, comprising the steps of: first, cropping the SAR image; second, calculating parameters of the segmentation image sparse model; third, calculating the weighted image by using the segmentation image sparse model; Four, select all parameter groups, and repeat step three; five, select a pixel-level fusion method or a decision-level fusion method to process the weighted test samples and identify the target. The invention uses the sparse representation model of the segmented image to separate possible occlusion areas and weakens the influence of occlusion on recognition by weighting the possible occlusion areas; adopts two fusion strategies of decision-level fusion and pixel-level fusion to avoid the segmentation image sparse model By combining the segmentation image sparse representation model with the fusion strategy, the occlusion information in the image can be judged and the interference caused by it can be judged, the dependence of the recognition method on the parameter value is reduced, and the occlusion target can be improved. recognition performance.

Description

technical field [0001] The invention belongs to the technical field of synthetic aperture radar (SAR) target recognition, and in particular relates to a SAR image fusion occlusion target recognition method based on sparse representation of segmented images. Background technique [0002] Synthetic aperture radar plays an increasingly important role in the fields of resource exploration, ocean observation, and battlefield surveillance due to its advantages of being unaffected by factors such as weather, time and light. Among these application fields of synthetic aperture radar, automatic target recognition of synthetic aperture radar has always been a hot and difficult research topic. Synthetic aperture radar target recognition technology can be divided into three types: template-based recognition, model-based recognition and machine learning-based recognition. [0003] Template-based recognition needs to build a complete template library. Although this method is simple in pr...

AI Technical Summary

Problems solved by technology

In the prior art, except for ISR, other methods do not take measures to eliminate occlusion in the processing process, but process the interference information caused by occlusion together with the target information

Interference information can affect the calculation results of these methods, leading to false identification of occluded targets

Although ISR attempts to eliminate the influence of occlusion by adding an error term in the sparse representation model, the solution method of the traditional sparse representation model will cause the final error term to accurately reflect the various occlusion situations.

Therefore, compared with the traditional SR method, although ISR can better identify occluded targets, there is still a lot of room for improvement in its recognition performance.

Images