A kind of image processing method and device based on Gaussian cloud transformation

Abstract

The invention discloses an image processing method and device based on Gaussian cloud transformation. The method includes: forming a gray value data sample set of the image by extracting the gray value of each pixel of the image; The number of crests of the frequency distribution of the value data sample set is obtained to form the number m of the initial Gaussian cloud; according to the number m, the gray value data samples are assembled into m Gaussian clouds; the ambiguities of the m Gaussian clouds are sequentially Degree of ambiguity is compared with the threshold value of ambiguity to determine whether there is a Gaussian cloud whose ambiguity is greater than the threshold of ambiguity; Gray value data samples are collected into a Gaussian cloud and Gaussian cloud ambiguity is compared with the ambiguity threshold until the ambiguity of all Gaussian clouds is less than or equal to the ambiguity threshold; output its ambiguity is less than or equal to the ambiguity threshold Degree threshold for all Gaussian clouds.

Description

technical field [0001] The invention relates to the field of image processing, in particular to a method for image processing using Gaussian cloud transformation and related devices. Background technique [0002] Granular computing is a method for studying and simulating human beings to represent, analyze and reason things from different granularities and levels, and it is an important direction in the research of intelligent information processing technology in artificial intelligence. The representation, generation, and selection of appropriate granularity in human cognition have always been a difficult problem for granular computing. [0003] Cloud model is a cognitive model based on probability theory to study qualitative and quantitative conversion. The cloud model uses three digital features, expectation, entropy and hyper-entropy, to characterize the connotation of a qualitative concept. Expectation is the data sample that best represents the concept. The uncertaint...

AI Technical Summary

Problems solved by technology

At present, since the Gaussian transformation determines which target area the pixel belongs to according to the probability value of each pixel in the image corresponding to each Gaussian distribution, when the image is segmented, there are often inconsistencies at the junction of the two areas. The natural sawtooth cannot reflect the uncertainty area between the target areas; and the Gaussian transformation needs to pre-specify the number of target areas, and cannot adaptively obtain the number of image targets according to the statistical characteristics of the pixel gray value

Images