An image sensor pixel array comprises a plurality of image pixel units to gather image information and a plurality of phase detection auto-focus (PDAF) pixel units to gather phase information. Each of the PDAF pixel units includes two of first image sensor pixels covered by two micro-lenses, respectively. Each of the image pixel units includes four of second image sensor pixels adjacent to each other, wherein each of the second image sensor pixels is covered by an individual micro-lens. A coating layer is disposed on the micro-lenses and forms a flattened surface across the whole image sensor pixel array. A PDAF micro-lens is formed on the coating layer to cover the first image sensor pixels.