A method of forming an image sensor device is disclosed. The method includes providing a substrate having sensor elements in a pixel region and having no sensor elements in a non-pixel region. The method further includes forming metal pillars over the pixel region and a metal shield layer over the non-pixel region. The metal pillars are disposed above spaces between adjacent sensor elements. The method further includes depositing a dielectric layer over the metal pillars and the metal shield layer; and etching the dielectric layer to form first and second trenches. The first trenches are formed over the pixel region and the second trenches are formed over the non-pixel region. Each of the first trenches aligns to a respective sensor element and is surrounded by the dielectric layer at its bottom and sidewall surfaces.