多通道信息融合的改进乘积规则

用传统的多通道信息融合乘积计算出的融合分值的动态范围比较小，融合效果不理想，本文在假定所有通道都能同时正常工作的情况下，提出多通道信息融合的改进乘积规则，基本思想是对原乘积规则中的概率密度和融合分值平滑，使得融合分值的动态范围增大，文中的两个实验表明，改进的乘积规则的识别率有明显提高，最后文中给出了在已知每个单一通道系统识别率的条件下，同理想融合方法集成出的系统识别率的上限计算公式。

The Improved Product Rule for Multimodal Information Fusion

The fusion scores calculated by the traditional product rule for multimodal information fusion have a narrow dynamic range,thus the fusion performance of this method is usually not ideal.Under the assumption that all the modes can work properly at the same time,this paper introduces the improved product rule for multimodal information fusion.The basic idea is to increase the dynamic range of fusion scores by smoothing the probability densities and fusion scores.Two experiments have shown that the recognition rate based on the improved product rule is increased.Finally,this paper presents a method for calculating the upper bound of recognition rate of the integrated system by an ideal fusion approach under the given recognition rate of each mode.