A tutorial on implementing a fast Gaussian blur in Java using two key optimizations: separable 1D filters and repeated box blurs with a sliding window. The standard 2D convolution runs in O(n·r²), but separating horizontal and vertical passes reduces it to O(n·r), and using a sliding-window moving average for box blurs further
Table of contents
1. Introduction2. Problem Overview3. The Fast Approximation Algorithm4. Solution5. Test6. ConclusionSort: