Fast Experimental Algorithm for 2D Otsu Thresholding

In this paper, we introduce a fast experimental algorithm for 2D Otsu thresholding that accelerates threshold computation for images, thereby improving image processing efficiency. Based on the standard Otsu algorithm, our enhanced implementation incorporates integral image techniques to precompute cumulative sums of pixel intensities and their squares, reducing time complexity from O(L^4) to O(L^2) where L is the intensity level. The algorithm maintains high precision by preserving the original Otsu's between-class variance maximization principle while optimizing histogram traversal through dynamic programming. Experimental results demonstrate that this approach reduces computation time by over 50% without sacrificing accuracy. For detailed implementation including pseudo-code showcasing the integral histogram calculation and variance computation loops, along with comprehensive experimental data, please refer to the relevant literature. Key functions include: - Precomputation of 2D integral histograms using cumulative sums - Efficient between-class variance calculation via precomputed statistical moments - Optimized threshold search through iterative updates of class probabilities