A locally adaptive silicon artificial retina with edge detection operation has been designed. Local brightness adaptation is achieved by normalizing each individual pixel photocurrent to the average photocurrent of the local adaptation area. The maximum photocurrent is estimated about 4nA with $100㎛^2$ photodiode area and the size of adaptation area is programmable from shift register addressing. In the adaptation area, edges are detected from subtracting output currents of two vertically neighboring pixels in the center of the area. The pixel circuit has been optimized by SPICE simulations and tested with MATLAB simulations. The designed pixel has 56.6㎛×56.4㎛ area and 5.97% Fill Factor. The Dynamic Range of retina is expected as 111dB from the SPICE simulations and the enhancement of edge detection in the dark region has been demonstrated from MATLAB simulations.