Compressive stereo cameras for computing disparity maps

Compressive imaging employs the direct measurement of object features and has been shown to offer both performance (e.g., improved reconstructed image fidelity) and cost (e.g., reduced number of measurements relative to the native dimensionality) advantages. We examine compressive imaging within a stereo vision application in which a traditional correspondence algorithm is used to find pixel disparity maps. Through simulation we show that compressive imaging provides sufficient image fidelity with 12.8× compression to compute disparity maps with less that 4.5% error on average at 0.5% relative measurement noise strength.