In this study, an image processing algorithm is developed for epiretinal implants. In this algorithm, it is presented an approach for adaptive frequency control that is one of the important parameters to generate stimuli. Standard 240x320 pixels video images in RGB format are processed as 10x10 pixel blocks. Consequently, it is designed to activate 24x32 pixels electrode matrix. As a retinal filter Gabor filter is preferred to DoG filter that is usually used in literature. The changes in pixels are modeled by using temporal filter. The results that belong to study and forward opinions are also presented.