Comparative evaluation of birth and death method and orthogonal matching pursuit in image analysis

Abstract

We present a new approach to extract predefined circular objects, such as trees and oil tanks, from aerial images. We consider a stochastic approach based on an object process also called marked point process. The objects represent trees or oil tanks which are modeled by their geometrical properties in the image. We first define a Gibbs density that takes into account both prior information and the data. The energy we defined is composed of two terms, one is a prior, penalizing overlaps between objects, and the other is a data term, which measures the suitability of an object in the image. After constructing the model and energy function, then the problem is reduced to an energy minimization problem. We sample the process to extract the configuration of objects minimizing the energy by a fast birth-and-death dynamics, leading to the total number of objects (trees or oil tanks in our case). In large images, this approach is much faster than manual counts and does not need any preprocessing or supervision of a user. We also adapted the algorithm to extract rectangular buildings in aerial images. We used the invariant color feature and shadow information in order to construct the birth map for buildings. Birth and death method is a faster algorithm compared to the previous methods. The drawback of the algorithm is that the running time doesn't change with the sparsity of the image. Thus, we have adapted orthogonal matching pursuit algorithm to estimate the number of objects in a fast manner in sparse images.