The probabilistic Ant Colony Optimization (ACO) approach is presented to solve the problem of designing an optimal route for hard combinatorial problems. The analysis presented in this paper focuses on haze removal for underwater images which have poor visibility due to presence of haze. It is found that most of the existing researchers have neglected many issues i.e. no technique is accurate for different kind of circumstances. The existing methods have neglected the use of ant colony optimization to reduce the noise and uneven illuminate problem. The main objective of this paper is to evaluate the performance of Ant colony optimization based haze removal over the available MIX-CLAHE (Contrast Limited Adaptive Histogram Equalization) technique. The experiment has clearly showed the effectiveness of the proposed technique over the available methods.