摘要
By combining fractal theory with D-S evidence theory, an algorithm based on the fusion of multi-fractal features is presented. Fractal features are extracted, and basic probability assignment function is designed. Comparison and simulation are performed on the new algorithm, the old algorithm based on single feature and the algorithm based on neural network. Results of the comparison and simulation illustrate that the new algorithm is feasible and valid.
提出了一种基于多分形特征融合的目标识别算法 .在此算法中 ,将分形理论与D S证据融合理论相结合 ,提取或构造了分形特征 ,设计了合理的概率分配函数 ,并对所提出的算法进行了仿真研究 ,并将此算法的识别结果与基于单分形特征的识别算法、基于神经网络的目标识别算法进行比较 ,结果表明本算法是可行的和有效的 .
