Block Inverse Hough Transform based on boundary trend prediction

A new block inverse Hough transform algorithm based on boundary trend prediction is proposed. Suppose there is only one main boundary, which can be expressed in several straight line segments. Several image blocks along the boundary are extracted. For each image block, the algorithm will use the IHT to reconstruct the pixels of main lines, which can best reflect the characteristics of the corresponding boundary. At last, all the main lines are connected to form the boundary of the whole image. The algorithm analysis indicates that the algorithm greatly reduces the memory requirement and computational complexity of standard inverse Hough transform (SIHT). The computational result shows that this new method is noise insensitive like HT and has precise reconstructive nature of IHT and that the prediction of boundary trend reduces the blindness of pixel reconstruction.