基于感知与信息熵融合的离散LOD模型智能组合(英文)

Intelligent Combination of Discrete LOD Model for 3D Visualization Based on Visual Perception and Information Entropy Fusion

针对图像信息熵单指标评价在三维场景离散LOD模型智能组合优化过程中容易造成可信度低的问题,提出了一种基于视觉感知与信息熵融合的离散LOD模型智能组合方法。分析了基于视觉感知的图像质量评价指标,构建了多指标融合的离散LOD模型智能组合框架,设计了与基于PSO的单指标离散LOD模型组合寻优对比实验。结合遗传算法的思想对粒子群算法进行改进,克服了在模型组合寻优过程中粒子群算法易于陷入局部最优的缺点。实验结果表明,本文方法能够设计出符合人类视觉感知特征的高可信度三维场景,与其他方法相比具有模型组合寻优效率高、无需人工交互的优点。

With respect to the problem that single-index evaluation of visual quality in the process of intelligent combination of discrete Level-of-Detail(LoD) model for three-dimensional(3D) visualization can easily result in low reliability. Human visual perception and information entropy based multi-index fusion intelligent combination of discrete Lo D model algorithm was proposed. Through analyzing the visual perception based image quality evaluation index, multi-index fusion evaluation intelligent combination of discrete Lo D model framework was built. Comparative experiment with single index model combination bases on PSO method was designed. Model combination optimizing process utilized modified Particle Swarm Optimization(PSO) method which was strengthened by Genetic Algorithm(GA) to avoid falling into local optimum. Experimental results demonstrate that the proposed method can design high reliability 3D visualization effect which is adapted to human visual perception characteristics, and outperforms other matching methods in designing efficiency and requires no user interaction.