Image Enhancement by Multiobjective Optimization and Bio-inspired Heuristics

In image analysis and computer vision applications the results precision and correctness depend on the quality of processed images. The most common parameter which subjectively defines the image quality is its contrast. In this paper an image contrast enhancement procedure based on multiobjective optimization is proposed. The contrast gain which has to be maximized and tone distortion which has to be minimized are used as optimization criteria. Because the histogram optimization is a high-dimensional problem, as optimization algorithm the usage of nature-inspired heuristics is proposed. Particularly, in the experiments presented in this paper, the Particle Swarming Optimization algorithm is used. Our contrast enhancement method proves to be superior to traditional techniques like histogram equalization in terms of contrast gain and tone distortion, both criteria being optimized.