Multimodal fusion perception has become increasingly prominent in unmanned systems, and the quantitative evaluation of data quality throughout the fusion process is now a crucial concern for ensuring the robustness and precision of perception algorithms. This study establishes a comprehensive and systematic metric system for evaluating data fusion quality, aiming to holistically represent the data quality across multiple dimensions during fusion. To validate the feasibility and practical value of this metric system, a digital measurement and evaluation device was developed. This device supports multimodal data acquisition and facilitates real-time processing and analysis of fusion perception algorithms. Leveraging a standardized multimodal dataset collected through this system, representative fusion strategies were selected for experimental comparison. The results demonstrate that the proposed metric system exhibits strong discriminative capability and adaptability, effectively supporting the quantitative assessment and performance analysis of multimodal fusion algorithms.