Mutated Kd-tree Importance Sampling

This paper describes a novel importance sampling method with applications in multimodal optimization. Based on initial results, the method seems suitable for real-time computer vision, and enables an efficient frame-by-frame global search with no initialization step. The method is based on importance sampling with adaptive subdivision, developed by Kajiya, Painter, and Sloan in the context of stochastic image rendering. The novelty of our method is that the importance of each kd-tree node is computed without the knowledge of its neighbours, which saves computing time when there’s a large number of samples and optimized variables. Our method can be considered a hybrid of importance sampling, genetic algorithms and evolution strategies.