In this exercise, you will reconstruct a 3D scene using dense epipolar matching. You will achieve this with the following steps: First, you will try to determine pixel disparity between a left and a right stereo frame using SSD matching on a disparity range. Second, you will apply some very simple heuristics to remove outliers. Third, you will backproject the matched pixels and triangulate the corresponding 3D point. Finally, you will use pose information of the frames to accumulate a global point cloud and visualize it in CloudCompare software