Computational model for using Gestalt Principles

The Gestalt Principles are thought to be the mental laws governing the organization of perceptual scenes. They are widely used in understanding human perception of different objects in a given image.

The main aim of this project is to design a computational model for segmenting an image based on gestalt principles.

The main motivation behind using Gestalts laws for segmenting image is the very simplicity of Gestalts laws. Since the identification of Gestalts laws by Wertheimer ¹, many persons like Kohler and Koffka have tried to state them quantitively and implement them for human like vision. I am using ² for reference purpose. While gestalts laws and their relative priority are generally dealt qualitativily, ² and ⁴ gives a solid quantitative interpretation for two rules using a distance function in probabilistic settings.

This project is continuation of previous work by Dipendra and Amit Barjatya⁵. They have used two laws of Gestalt to segment images,

1. Law of Color Constancy
2. Law of Continuity

Continuing on the line of previous project I am adding one more law Gestalt's law of symmetry. Dipendra and Amit have implemented two laws but their law of continuity, owning to its complications is implemented for some specific situations only. I am also looking forword to improve this implementation. Since the previous work is done on C#, I will continue with it. Similarily the platform for code development will be Visual Studio 2012 added with Computer Vision libraries from Emgu CV. I will use their database for first two laws and will be construcing my own image database using MS Paint for all laws.