Vector quantization (VQ) plays an important role in data compression and it has been successfully used in image compression. In VQ, minimization of Mean Square Error (MSE) between code book vectors and training vectors is a non-linear problem. Traditional LBG types of algorithms used for designing the codebooks for vector quantizer converge to a local minimum, which depends on the initial code book. Genetic algorithms (GAs) are a powerful set of global search techniques that have been shown to produce very good results on a wide class of problems. GAs are capable of exploring and exploiting promising regions of the search space. The genetic algorithm can be applied to generate a better codebook that approaches the global optimal solution of vector quantization. In this paper we present a new approach to image compression based on genetic algorithm for vector quantizer. We also propose a composite crossover operator for generating a codebook. The effectiveness of the proposed crossover operator on the design of a codebook of genetic algorithm based vector quantization is analyzed. Simulations indicate that vector quantization based on genetic algorithm has better performance in designing the optimal codebook for vector quantizer than conventional LBG algorithm. The result also indicate that the performance of the codebook is substantially improved by using the proposed cross over operator. The Peak Signal to Noise Ratio (PSNR) is used as an objective measure of reconstructed image quality.