Image just noticeable distortion threshold estimation method based on texture masking effect

Abstract

The invention discloses an image just noticeable distortion threshold estimation method based on a texture masking effect, which comprises the following steps of converting a grayscale image of an original image into a double-precision floating point type image, calculating fractal dimension of each image block in the image by adopting a differential box counting method, and taking the fractal dimension as a weight for measuring the texture roughness of the image block, obtaining a weight graph; calculating the texture complexity of each pixel point in the grayscale image according to the weight image, obtaining a texture masking effect according to the texture complexity and the brightness contrast, and obtaining a contrast masking effect according to the brightness contrast; extracting an object edge; according to the texture masking effect, the contrast masking effect and the binary image corresponding to the object edge, obtaining a space masking effect of each pixel point in the gray level image; obtaining an initial just noticeable distortion threshold according to the spatial masking effect and the brightness adaptability, and obtaining a final just noticeable distortion threshold in combination with the saliency map of the grayscale image; the method has the advantage of high just-noticeable distortion threshold estimation accuracy.

Description

technical field [0001] The invention relates to an image just noticeable distortion (Just Noticeable Difference, JND) threshold estimation method, in particular to a texture masking effect-based image just noticeable distortion threshold estimation method. Background technique [0002] With the emergence of different social software, people are more willing to upload their own life and entertainment photos or videos on these social software or networks to share their pleasures with others. This involves the image encoding task, how to hide useless and retain effective information is the key to saving network resources, and the JND threshold estimation technology is just for this purpose. In the past ten years, many JND threshold estimation methods have been proposed. However, most of the methods cannot estimate the JND threshold of the image very accurately. For example, some methods only consider the influence of brightness on the JND threshold of the image. , although som...

AI Technical Summary

Problems solved by technology

According to the characteristics of the Human Visual System (HVS), compared with areas with relatively flat textures, HVS has a higher tolerance for distortion in areas with rough textures. The JND threshold of these areas should be higher. However, the image JND threshold estimation method proposed in the past ignores this point, resulting in inaccurate estimation of the image JND threshold

Images