Disparity map algorithm using census transform and hierarchical segment-tree from stereo image

This article presents a disparity map algorithm to improve the depth map estimation based on Census Transform and hierarchical segment tree on each block. Essentially, the framework applied is based on reconstructed through a series of consecutive techniques. The process produces the disparity map or depth map as a result. To generate a disparity or depth map, a four-stage taxonomy has been used. Technically, the generated map contains essential information that will be used in a variety of usage such as drone navigation, autonomous vehicle exploration and navigation, virtual reality and surface reconstruction. To produce an appropriate depth map, the structure need resilient to low texture, bare colour, and repetitive pattern regions in the input stereo image. The stereo matching algorithm presented in this study comprises of four steps: Cost Computation, Cost Aggregation, Optimization, and Post-Processing, all of which will refine the final disparity map. Based on the research and results analysed using the existing benchmarking evaluation method which is Middlebury Benchmark, the disparity produced indicates a relatively low inaccuracy of 5.61% for nonocc error and 8.92% for all error pixels. On average, it outperforms and has a greater impact or equivalent to other proposed work.