Codebook design in Vector Quantization is a combinatorial optimization problem that requires balancing reconstruction distortion and compression rate. Multi-Objective Evolutionary Algorithms (MOEAs) enable the exploration of this trade-off, but their performance is often limited by insufficient exploitation in high-dimensional search spaces. This work proposes a decomposition-based hybrid MOEA that integrates a Micro-Genetic Algorithm (μGA) as a local search operator. The method partitions the problem into scalar subproblems defined by weight vectors, enabling targeted refinement of candidate solutions. The approach is evaluated on the design of the codebook for lossy image compression using the Kodak dataset. Results, assessed through standard multiobjective and signal quality indicators, show statistically significant improvements in convergence and diversity compared to state-of-the-art MOEAs.