Systems and methods for compressive image sensor techniques based on sparse measurement matrices are disclosed. A method to perform compressive sensing (CS) measurement operations for image sensors limits pixel summation to be within neighboring pixels and hence dramatically simplifies CS image sensor circuits and reduces their power consumption while providing better image quality compared to conventional random measurement matrix based methods. A sparse measurement matrix is applied to pixel data to generate a desired number of summation groups, each summation group consisting of outputs from an equal number of pixel cells. Each pair of summation groups contains the same number of shared outputs from pixel cells. From the summation groups, an image captured by the pixel cells is recovered.