Decoding brain activity occurs in response to an external stimulus is a great problem and the first step to analysis, classification, reconstruction of the external stimulus which will lead to some marvelous application like reconstructing the visual images from the mind of a person in comma or identifying an object in the head of someone that has some psychological problem in the medical field, and criminal image reconstruction instead of forensic artists in security problems or instant design in marketing, etc...
The problem with the brain data analysis is that the brain data like fMRI which we used here is very complicated. also acquiring the data is difficult and requires a lot of resources and time and this is why the fMRI datasets size is small. The data is divided into 3 subjects and each subject data consists of fMRI responses to 3 types of images(letters,artificial,natural). In this work, we focused on 3 main targets analysis and classification of the difference between subjects and types of images, identification of objects presented in the natural images, and finally reconstruction of the images from fMRI data.