No-reference tone mapping image quality assessment method based on multi-feature fusion

Abstract

The invention provides a non-reference tone mapping image quality evaluation method based on multi-feature fusion. The method is mainly divided into feature extraction stage and training regression stage. In the feature extraction stage, the features of the image are extracted in three fields, the entropy feature of the image is extracted in the pixel field, the texture feature based on the gray level co-occurrence matrix and the statistical feature of the natural scene; in the blur field, we use the local phase consistency to evaluate the overall blur In the field of color, we convert the image to the opposite color space, and measure the overall color information and the contrast information of each channel. Finally, the machine learning method is used to predict the quality of the tone-mapped image. The algorithm proposed by the invention can accurately and effectively predict the quality of the tone-mapped image.

Description

technical field [0001] The present invention relates to the technical field of tone mapping image quality evaluation, and is an image quality evaluation method based on feature extraction, in particular, it relates to a non-reference tone mapping image quality evaluation method based on multi-feature fusion. Background technique [0002] High Dynamic Range (HDR) images can accurately show the difference in brightness from gloomy starlight to bright sunlight (10 -3 cd / m 2 to 10 5 cd / m 2 ), which can bring viewers a more realistic and rich visual experience. However, existing image processing systems mainly use traditional 8-bit low dynamic range (Low Dynamic Range, LDR) display devices. In order to make the HDR image backward compatible with the LDR display device, a tone-mapping operator (Tone-Mapping Operator, TMO) for converting the HDR image into the LDR image is proposed. Ideally, the tone-mapped image (Tone-Mapped Image, TMI) should retain the original structure an...

AI Technical Summary

Problems solved by technology

[0004] 1. The current algorithm only considers certain single-category features and does not start from multiple perspectives, so the prediction accuracy is poor;

[0005] 2. The current algorithm only evaluates image quality from the perspective of the human visual system, and does not take into account the unique characteristics of TMI images, such as the unique halo effect of TMI images

Images