Efficient denoising-based visual image reconstruction method

Abstract

The invention discloses an efficient denoising-based visual image reconstruction method, and belongs to the technical field of biomedical image mode identification. The method comprises the steps of first performing dimension reduction processing on an eigenvector, then performing image conversion processing on a contrast graph of stimulative images of training samples based on corresponding local image bases of different scales, and establishing corresponding classifiers; obtaining multiple groups of first prediction tags by classifying and predicting the training samples, and then obtaining an associated coefficient of each pixel point of a visual image; and inputting a to-be-processed fMRI (function Magnetic Resonance Imaging) signal, obtaining second prediction tags corresponding to different local image bases, and obtaining a reconstruction tag of a current pixel point by weighted summation of the associated coefficient of the same pixel point and the second prediction tags, thereby obtaining a reconstructed visual image. By implementing the method, the visual image can be reconstructed and the noise of the reconstructed image is reduced.

Description

technical field [0001] The method belongs to the technical field of biomedical image pattern recognition, and specifically relates to a visual image reconstruction method of a functional magnetic resonance image. Background technique [0002] The human brain has the advantages of high efficiency, robustness and anti-noise in processing complex visual information. Visual information is one of the most important ways for human beings to perceive and understand the external world. The brain has certain commonalities in the processing of different types of natural scenes. For example, the temporal ventral parahippocampal region is a brain functional area that is closely related to the processing of scene information, but the processing of scene information also depends on multiple visual brain areas. Brain function representation Visual information is highly complex. So far, we still know little about the brain area and its encoding mechanism that process complex natural scene...

AI Technical Summary

Problems solved by technology

Dating back to the 1990s, some researchers used external lacquer nerve signals to reconstruct visual scenes. Due to a series of limitations in social production technology and process technology at that time, although visual image reconstruction has made great progress. It has a certain effect, but the reconstruction accuracy is far from the ideal result, and it has traumatic damage to the subjects

With the rise and development of magnetic resonance technology, it is possible to study the brain completely non-invasively, such as using the visual reconstruction method of sparse multinomial logistic regression (SMLR). Although this method improves the existing reconstruction effect, the reconstruction result Contains a lot of noise

Images