Research on inference of visual images from incomplete neural activity data in primary visual cortex

点击翻译按钮获取中文摘要

In the research process, we found that super resolution technique, which infers a high resolution image from low resolution images, is more substantial and general technique of the inference of visual images from visual cortex data. Afterward, we researched the super resolution and succeeded to develop the best method in low noise level situation. The method can analytically and approximately determine the optimal estimator over the original image model of the compound Markov random field prior. Such problem had been an unsolved problem for a long time.