Research On Theory And Method Of Visual Reconstruction Of Brain Decode

The visual system is one of the important senses of human access to external environmental information.Human visual information processing is efficient,anti-noise and robust,especially in the recognition and processing of natural scenes.Behind the amazing visual information processing ability of the brain,what kind of neural mechanism is hidden? This is a fascinating problem for countless scientists,interpretation of human brain vision cognitive fast coding and decoding mechanism has been the most cutting-edge and challenging brain science direction.With the gradual clarity of the brain coding mechanism and the rapid development of brain imaging techniques and pattern recognition methods,scientists began to think about more complex and deeper problems-brain decoding.Whether the brain function of the signal can be reconstructed,to calculate the brain perceived visual information,such as text and images? Since 2005,some well-known international scientists and laboratories began to explore the brain decoding methods and techniques,in recent years has made great progress.Miyawaki established a method and technique of reconstruction of brain signal visual images based on candidate categories using early visual cortex retinal topology mapping.However,the reconstructed images of his method have great noise and the efficiency of reconstruction low.In view of this shortcoming,this paper proposes a Bayesian reconstruction method based on F-score feature selection to improve the accuracy and efficiency of image reconstruction.To achieve the reconstruction of visual images,the need for the visual cortex subregion of the brain positioning,so this paper carried out the retina topology mapping magnetic resonance experiments.The experiment uses the "wedge" and "ring" stimuli to induce the visual cortex of the brain,using the generalized linear model(GLM)to find the voxels in the visual cortex,calculate the polarities of the activated voxels and the eccentricity values,Polar angle map and eccentricity map.According to the polar angle map and the eccentricity map,the division of the primary visual cortex subarea is achieved,and then all voxel positions in the V1,V2,V3 regions are located.For the study of visual image reconstruction,this paper establishes a Bayesian visual image reconstruction model based on F-score feature selection algorithm.The model uses the F-score feature selection method to select the voxels with good reconstructing effect,removes the voxels that are not related to the stimulus image,and finally reduces the noise of the reconstructed image.The model also combines the low complexity Bayesian method to improve the efficiency of visual image reconstruction.