A reference-free stereo image quality assessment method based on fusion images

Abstract

The invention belongs to the field of image processing and aims to propose a no-reference stereoscopic image quality evaluation method, which is more in line with the characteristics of human eyes and maintains good consistency with human subjective perception. For this reason, the technical solution adopted by the present invention is to fuse the left and right views of the stereo image based on the no-reference stereo image quality evaluation method based on the fused image, and use the method of overlapping and cutting to send it to the neural network Alexnet for migration learning training, and the predicted The quality of the stereoscopic image, and finally use the visual salient characteristics of the human eye to weight the fused image. The invention is mainly applied to image processing occasions.

Description

technical field [0001] The invention belongs to the field of image processing, and relates to the improvement and optimization of a stereoscopic image quality evaluation method, and the application of visual salience in the human visual system in the objective evaluation of stereoscopic image quality. Specifically, it involves a no-reference stereo image quality assessment method based on fused images. Background technique [0002] In recent years, with the development of multimedia technology, more and more attention has been paid to stereoscopic images, and the quality of stereoscopic images will be degraded in the process of collection, compression, transmission, display, etc., and the quality of stereoscopic images will directly affect people's Visual perception, therefore, how to effectively evaluate the quality of stereoscopic images becomes one of the key issues in the field of stereoscopic image processing and computer vision. [0003] Since the subjective quality e...

AI Technical Summary

Problems solved by technology

However, the current understanding of the human visual system is still very limited, and traditional methods are difficult to fully reflect human visual perception of stereoscopic images. Therefore, researchers use neural networks that can simulate the human brain to evaluate the quality of stereoscopic images. Literature [6] A stereoscopic image quality assessment method based on support vector machine (SVM) is proposed

However, traditional machine learning methods need to manually select stereoscopic image features, and the selected features may not fully reflect the quality of stereoscopic images, which limits the development of stereoscopic image quality technology.

Images