论文部分内容阅读
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。
Aiming at the problem that single entropy evaluation of image entropy easily leads to low credibility in the process of intelligent LOD model discrete three-dimensional (3D) LOD modeling, a novel LOD intelligent combination method based on visual perception and information entropy fusion is proposed. The evaluation index of image quality based on visual perception is analyzed, and the intelligent combinatorial framework of discrete LOD model with multi-index fusion is constructed. The optimal combination contrast experiment with the single-index discrete LOD model based on PSO is designed. Combining with the idea of genetic algorithm, PSO is improved, which overcomes the shortcoming of particle swarm optimization prone to local optimum in the process of model combination optimization. The experimental results show that this method can design high-confidence three-dimensional scene which is in line with human visual perception features and has the advantages of high efficiency and no human interaction compared with other methods.