Light-field microscopy enables real-time volumetric imaging, offering substantial advantages for real-time fluorescence imaging. However, when applied to multi-color imaging, positional cross-talk between different fluorescent signals in the object space reduces reconstruction accuracy. Additionally, imaging each color through sequential excitation of fluorophores significantly compromises imaging speed. In this paper, an image response-assisted volumetric reconstruction method that unmixes multi-color fluorescence light-field images using pixel support derived from the light-field imaging response is proposed. This approach enables simultaneous multi-color imaging with significantly improved volumetric reconstruction accuracy. The correctness and effectiveness of the proposed method are validated through both simulations and experiments. The root-mean-square-error of multi-color volumetric reconstruction is reduced by 37.87 % on average compared with the simultaneous multi-color reconstruction methods obtained by simply combining single-pixel spectrum conversion methods and linear unmixing method in rapid-moving micro-particle observation, showcasing high accuracy simultaneous multi-color imaging performance. Volumetric imaging of motor neurons and whole-body cells of live dual-color zebrafish larvae at 20 Hz demonstrates the ability to be applied to real biomedical imaging.