复杂特征图像视觉显著性训练数学模型仿真

Mathematical Model Simulation of Visual Saliency Training for Complex Feature Images

对于复杂特征图像,传统的视觉显著性训练模型在处理目标图像边缘区域时,很难保留完整的结构信息,训练结果的真实性有待提高。针对上述问题,构建新的复杂特征图像视觉显著性训练数学模型。根据图像的全局对比特征以及亮度、饱和度、色调等属性,分别计算出对应的显著性参数。建立约束函数,整合计算得到的各个显著性参数,设计与显著性参数相对应的仿真参数。并根据整合的显著性设定新参数实现参数的统一管理。构建仿真联盟结构,实现数学模型的整体仿真。实验结果表明,设计的视觉显著性训练数学模型信息完整,仿真偏差小,显著性明显,整体真实性更高,与传统方法相比,实用性更强。

For complex feature images,traditional models were difficult to retain complete structural information when processing the edge region of the target image,and the authenticity of training results should be improved.Therefore,a mathematical model for visual saliency training of complex feature images was constructed.According to the attributes of the image such as global contrast characteristics,brightness,saturation and hue,significance parameters were calculated respectively.Moreover,constraint functions were established and integrated,and then significance parameters were obtained.Furthermore,the simulation parameters corresponding to the significance parameters were designed.According to the significance,new parameters were set to achieve the unified management of parameters.Finally,a co-simulation structure was constructed to realize the overall simulation of the mathematical model.Experimental results show that the designed model is significant with the features of complete information,small simulation error,higher authenticity and better practicability.