Single image super-resolution method based on joint contour enhancement and denoising statistical prior

Abstract

The invention discloses a single image super-resolution method combining contour enhancement and denoising statistical prior. It mainly includes the following steps: introduce a priori based on deep learning and a continuity mechanism, and obtain a single image super-resolution reconstruction framework based on the improved SBI algorithm; construct and train the contour enhancement network PENet and the denoising statistical prior network DSPNet; Construct the image contour enhancement prior PEP, and apply it to the image restoration inverse sub-problem; use TFOCS technology to optimize the image restoration inverse sub-problem described in step 1; calculate the noise level σ k ; Construct image statistical prior DSP, and apply it to the denoising sub-problem; update parameters; perform iterative reconstruction, and output the final super-resolution reconstruction result. The single image super-resolution reconstruction method described in the present invention can obtain good subjective and objective effects, and has a fast operation speed. Therefore, the present invention is an effective single image super-resolution reconstruction method.

Description

technical field [0001] The invention relates to an image resolution improvement technology, in particular to a single image super-resolution method combined with contour enhancement and denoising statistical prior, belonging to the field of image processing. Background technique [0002] Image super-resolution reconstruction techniques use a single or a set of low-resolution images (sequences) to produce high-quality, high-resolution images. The application fields of image super-resolution reconstruction are extremely broad, and there are important application prospects in military, medical, public security, computer vision and other aspects. In the field of computer vision, image super-resolution reconstruction technology can transform the image to the identification level to improve the recognition ability and recognition accuracy of the image, and provide rich image information for the subsequent analysis process. [0003] Single image super-resolution reconstruction met...

AI Technical Summary

Problems solved by technology

However, it is very challenging to significantly improve the super-resolution reconstruction effect by simply changing the traditional explicit prior term form or designing a deeper neural network structure.

Images