Grayscale image fusion method

Abstract

The invention provides a grayscale image fusion method. A global image and a local image are generated on a low-dynamic image by using a low-rank decomposition method, different weight maps of two decomposed images are constructed according to low-illumination imaging features, and the block effect after fusion is reduced, so that the visual effect of the image after fusion better conforms to the consistency of human eyes; the decomposed low-rank image and saliency image are combined in a Laplacian space by using a Gaussian pyramid weight factor to obtain a fused low-rank image and saliency image; and finally, a high-dynamic image is recreated by superposing the low-rank image and the saliency image. According to the method, detail features of the dark area and the highlight area of the image are kept, and the dynamic range of the image is widened.

Description

technical field [0001] The invention belongs to the technical field of high dynamic image synthesis, and in particular relates to a grayscale image fusion method based on low-rank decomposition. Background technique [0002] Common CCD sensors have a dynamic range of about 10 3 , much smaller than the dynamic range of real scenes. Additionally, images captured in low-light conditions tend to have poor visibility. Lost image details are difficult to recover due to limited dynamic range. Most of the existing multi-exposure fusion methods are aimed at color images, and the original method of constructing weight maps cannot obtain the detailed information of the over-bright and over-dark areas, resulting in low quality of the fused image. At the same time, low-illumination images have low contrast, and the fusion weight map is not constructed, resulting in uneven edge transition of the synthesized image and block effects in the image. Contents of the invention [0003] In ...

AI Technical Summary

Problems solved by technology

Also, images captured in low-light conditions tend to have poor visibility

Lost image details are hard to recover due to limited dynamic range

Most of the existing multi-exposure fusion methods are aimed at color images. The original method of constructing the weight map cannot obtain the detailed information of the over-bright and over-dark areas, resulting in low quality of the fused image.

At the same time, the contrast of low-light images is low, and the fusion weight map is not used to construct the fusion weight map, resulting in uneven edge transition of the synthesized image and block effect in the image

Images