项目名称: 基于尺度相关感知误差测度近似全局优化的数字图像半色调方法研究
项目编号: No.61302173
项目类型: 青年科学基金项目
立项/批准年度: 2014
项目学科: 无线电电子学、电信技术
项目作者: 何自芬
作者单位: 昆明理工大学
项目金额: 24万元
中文摘要: 动态环境下多尺度数字图像半色调相对静态环境下单一尺度半色调更具实际意义和挑战性,其中多尺度感知误差测度融合算法的鲁棒性和优化结果的一致性是实现高质量半色调的关键约束。现有基于感知误差测度优化的半色调方法存在缺乏尺度交互以及收敛于局部极值的局限,本项目研究基于尺度相关感知误差测度近似全局优化的数字图像半色调方法,主要内容包括:1)多尺度感知误差测度模型框架与尺度相关感知误差测度建模理论,来融合误差测度的尺度内聚类性和跨尺度持续性统计分布规律,以保证融合算法鲁棒性;2)研究误差测度凸对偶转化策略与近似全局优化理论,来计算确定近似范围内的误差测度全局优化,以保证优化结果一致性;3)在典型动态环境下进行算法有效性验证实验,并与分级混合直接二值搜索算法对比,做出定量评价。本项目旨在探索多尺度感知误差测度鲁棒融合算法,以解决多尺度误差测度优化结果不一致难题,为多尺度误差测度全局优化提供理论指导和依据。
中文关键词: 数字半色调;多尺度;尺度相关感知误差测度函数;全局优化;
英文摘要: Multiscale digital halftoning under dynamic environment is a more challenging work than single scale halftone under static environment ones. The key problem to achieve the high-quality halftone are the robustness of the integration algorithm of multiscale perceived errrors measure and the consistence of the optimize results. The existing halftoning algorithm are the lack of the scale-interactive and only guaranteed to produce a local optimum. Themain contents include: 1)The model of multiscale perceived error metric and the theory of scale-related perceived error metric are to fuse the statistical distribution of the error metric of the scale of clustering and cross-scale persistent, which to achieve the robustness of the algorithm. 2) We apply the dual transformation error metric convex global optimization strategies and approximation theory to calculate the approximate range of error and determine the measure of global optimization, which to ensure consistency of optimized results. 3) Experimental results indicate the effectiveness of the algorithm under dynamic environment. Compared with the DBS algorithm and quantitative evaluation. This project aims to explore multiscale measure of perceived error robust fusion algorithm, for the purposed of unconsistency difficulties in optimized results of multiscale pe
英文关键词: Digital halftoning;Multi-scale;Scale-related perceived error metric function;Global optimization;