StereoBM

StereoBM, also known as the Block Matching algorithm for stereo vision, is a widely used method for computing the disparity map from a pair of stereo images. The fundamental principle behind StereoBM is to find corresponding pixels in the left and right images that represent the same point in 3D space. This correspondence is established by searching for a matching block of pixels in the right image for each block of pixels in the left image.

The algorithm operates by defining a search window in the left image and a corresponding search window in the right image. For each pixel in the left image, the algorithm slides a rectangular block of pixels (e.g., 15x15) across a predefined horizontal search range in the right image. At each position, a similarity metric is computed between the block in the left image and the current block in the right image. Common similarity metrics include Sum of Absolute Differences (SAD) and Sum of Squared Differences (SSD). The disparity is then assigned as the horizontal shift that yields the minimum difference between the two blocks.

To improve robustness and handle occlusions, StereoBM incorporates several post-processing steps. These often include a uniqueness constraint, which ensures that a pixel in the left image is matched to only one unique pixel in the right image. A sub-pixel refinement step can also be applied to achieve a more accurate disparity estimation. StereoBM is computationally efficient and well-suited for real-time applications, making it a popular choice in fields such as robotics, augmented reality, and autonomous driving. However, its performance can degrade in textureless regions or areas with repetitive patterns.