摘要
针对虚拟海洋战场如何绘制得更加符合观察者视觉注意特征问题,提出了一种能够智能选择多分辨率模型层级的基于视觉熵的迭代优化方法。在分析基于谱残差法的显著图生成原理基础上,设计了一种根据场景渲染图计算视觉熵的新方法,将求解得到的视觉熵作为适应度值,采用改进的人工鱼群算法对各分辨率层级的模型组合进行搜索、匹配和重组,经过不同分辨率模型的迭代修正后再次绘制的场景达到更加逼真的视景仿真效果。实验结果表明,该方法生成的三维虚拟海洋战场在人类通常关注的位置具有更多的细节,场景真实感更强。
In view of how to draw the 3 D virtual sea battlefield more in line with the observer’s visual attention feature,a new intelligent resolution method is proposed based on visual entropy and iterative optimization.Firstly,the principle of saliency map generation based on spectral residual method was analyzed,and a new computing method for visual entropy using rendering image was put forward.Then,the improved artificial fish swarm algorithm,with visual entropy as the fitness value,was adopted for searching,matching and reorganizing the models combinations at each resolution level.Finally,a scene with more realistic visual effects was obtained after several times of iteration.The experimental results show that the virtual sea battlefield generated by the method contains more model details and the scene is more realistic.
作者
徐建华
顾浩
汪小东
曾艳阳
XU Jian-hua;GU Hao;WANG Xiao-dong;ZENG Yan-yang(Marine College,Northwestern Polytechnical University,Xi'an Shanxi 710072,China;National Key Laboratory of Underwater Information Process and Control,Xi’an Shanxi 710072,China;College of Computer Science and Engineering,Sanjiang University,Nanjing Jinagsu 210012,China;College of Computer Science and Technology of Henan Polytechnic University,Jiaozuo Henan 454000,China)
出处
《计算机仿真》
北大核心
2019年第3期312-316,共5页
Computer Simulation
基金
国家自然科学基金(61503124)
江苏高校"青蓝工程"资助
关键词
视觉熵
谱残差
多分辨率
人工鱼群算法
海洋战场
Visual entropy
Spectral residual
Multi-resolution
Artificial fish algorithm
Sea battlefield
