Abstract—The paper presents new method to improve computational performance by introducing the mutual spatial feature in order to make strong visual cue in image parsing problem based on non-parametric model. This feature models the spatial context and mutual information in our previous study [1] to enhance accuracy and performance of image parsing problem in calculating the probability of co-occurrence objects. The experimental results based on Matlab programming language using SIFTFlow and Barcelona datasets showed that the mutual-spatial feature is promising to refine image parsing problem.

Index Terms—image parsing, mutual-spatial, Matlab, SIFTFlow

Cite: Thi Ly Vu and Chong Ho Lee, "Robust Method to Compute Mutual-Spatial Feature for Image Parsing Problem" Journal of Image and Graphics, Vol. 2, No. 1, pp. 54-58, June 2014. doi: 10.12720/joig.2.1.54-58