In recent years, there has been growing interest in reversible data hiding technology based on Multiple Histogram Modification (MHM). Existing methods for reversible data hiding based on multiple histograms are primarily designed for grayscale images. However, color images more widely used in daily life than gray images. Therefore, we propose a novel MHM-based reversible data hiding method for color images in this study. Leveraging the similar gradient structures among the Red, Green, and Blue (RGB) channels of color images and the correlation between prediction errors, we introduce a new metric measure the local pixel complexity. Multiple histograms are generated based on this complexity, and genetic algorithm is employed to adaptively select embedding points according to the characteristics of each histogram. Specifically, we present a channel-histogram synergistic payload partitioning strategy, which first adaptively divides the data payload based on the characteristics of each channel, rather than distributing it equally. Subsequent redistribution across histograms is performed by recalculating segmentation thresholds using adjusted cumulative distribution functions, enabling dynamic allocation. Comparative analysis with existing methods demonstrates that the proposed method achieves superior performance in terms of both visual quality and algorithmic complexity.