Segmentation of images is a very challenging problem due to the presence of noise in the images and its widespread usage and applications. In this paper we proposed the GABC-Genetic Artificial Bee Colony Algorithm which is a hybrid concept of Artificial Bee Colony and Genetic Algorithm used for solving segmentation problem. Here we add Genetic Algorithm with ABC Algorithm which improves the solution space. In GABC, threshold estimation is regarded as search procedure which finds an appropriate value in a continuous grey scale interval. In our proposed methodology, optimal threshold is searched with the help of ABC algorithm and the GA updates the solution space. For finding efficient fitness function for ABC algorithm the original image is decomposed using discrete wavelet transform after the definition of grey number in the Grey theory. Now the approximation image and gradient image is reconstructed with low frequency coefficients and high frequency coefficients respectively. Then a filtered image is produced with noise reduction to the approximation image. Therefore a co-occurrence matrix is constructed based on filtered image and gradient image. Then we define improved two-dimensional grey entropy which serves as the fitness function for GABC. And finally optimal threshold is rapidly discovered by the behavior of ABC operators in honey bee colony. Here the two operators of ABC, employed bees and onlooker bees are extended with genetic processes, crossover and mutation. Initially set of schedules are generated by the GABC algorithm which has to be evaluated against constraints and infeasible solutions has been resolved to feasible ones. And finally the algorithm iteratively improves the initial schedules until the termination condition is met. Artificial Bee Colony algorithm is used for global search strategy and the Genetic algorithm is used for local search strategy. The hybrid approach of GABC model can improve the result of ABC. The experimental result reveals that our GABC model can give a near to approximation and improved result and reach a broader domain in search space. It also improves both the computation time and precision and better than GA, ABC, PSO and AFS.