Retinex image enhancement algorithm based on Laplace pyramid reconstruction

Abstract

The invention discloses an image enhancement algorithm based on adaptive Retinex and wavelet fusion. The method comprises the following steps: firstly, performing illumination estimation on a picture by using a self-adaptive Retinex algorithm, and enhancing a reflection component to obtain a self-adaptive image; secondly, respectively sharpening in a frequency domain and a space domain on the basis of obtaining the self-adaptive image, namely respectively performing Gaussian high-pass filtering sharpening and improved space operator sharpening on the self-adaptive image to obtain a frequency domain sharpening image and a space domain sharpening image; and finally, dividing the adaptive image, the frequency domain enhanced image and the spatial domain sharpened image into high frequency and low frequency through wavelet transform, and respectively performing weighted fusion to obtain a final image. A relatively good enhancement effect is achieved on common images with non-uniform illumination and low-illumination images in practical application, and the algorithm can be applied to more occasions.

Description

technical field [0001] The invention relates to the technical field of image enhancement, including Retinex algorithm, Fourier transform, Gaussian high-pass filter, sharpening algorithm and wavelet transform. Background technique [0002] With the progress of society and the development of science and technology, digital image processing technology has been gradually applied in other disciplines and plays an important role in people's daily life, but the images collected under the influence of bad weather will appear contrast Factors such as low image quality and loss of details seriously affect the transmission of information, so the research on image enhancement has important significance. [0003] At present, image enhancement can be divided into two categories: spatial domain algorithms and frequency domain algorithms; [0004] The current traditional image enhancement methods include: linear transformation method, histogram equalization method, single-scale Retinex alg...

AI Technical Summary

Problems solved by technology

[0007] Image distortion, the fusion algorithm cannot extract effective information from two or more images, but introduces noise

[0008] It is greatly affected by low-quality images. When there are low-quality images in the fused image, the fusion effect will be affected by the low-quality images, which will reduce the image quality.

Images