Joint segmentation and image interpretation

Abstract

The problem of image interpretation is formulated in the framework of modular integration and multiresolution. The formulation essentially involves the concept of reductionism and multiresolution, where the image interpretation task is broken down into simpler subtasks of segmentation and interpretation. Moreover, instead of solving the vision task at the finest resolution Ω, we solve the synergetically coupled vision subtasks at coarser resolutions $\text{Ω}\text{−ξ}$ for $\text{Ω}\text{⩾ξ>0}$ and use the results obtained at resolution $\text{(}\text{Ω}\text{−ξ)}$ to solve the vision task at resolution $\text{(}\text{Ω}\text{−ξ+1)}$. We present a solution to the joint segmentation and interpretation problem in the proposed framework. For the interpretation part we exploit the Markov random field (MRF) based image interpretation scheme developed by Modestino and Zhang [IEEE Trans Pattern Anal. Mach. Intell. pp. 606–615 (1992)]. Experimental results on both indoor and outdoor images are presented to validate the proposed framework.