基于数字图像半色调技术的基本理论，本文对半色调算法进行了系统的分类，讨论了各种误差扩散算法和结构敏感的半色调算法的理论框架，分析了半色调算法中最前沿的结构敏感的误差扩散算法，并给出了具体实现步骤。通过对比处理结果，进一步肯定了该算法在数字图像半色调处理中的优越性。文章首先简要介绍了数字图像半色调技术以及半色调算法分类。接着讨论了各种误差扩散算法，指出其核心是阈值和误差过滤器的设计；讨论了结构敏感的半色调算法，指出其核心是动态全局优化。进而从局部频率分析、阈值调制、误差扩散系数调制和参数预校正四个方面分析了结构敏感的误差扩散算法，并给出了其对应的最佳窗口大小和处理器设计步骤。最后通过对各种算法的处理结果对比表明，结构敏感的误差扩散算法结合了其他算法优点，处理速度快，质量好，具有更加广泛的应用范围。

Based upon basic principles of digital halftoning, this paper systematically classifies the halftoning algorithms, discusses the theoretical frameworks of various error diffusion algorithms and the structure-aware halftoning algorithm, analyzes the up-to-date structure-aware error diffusion algorithm, and gives its implementation steps. By comparing the results, the structure-aware error diffusion algorithm shows its superiority. In this paper, digital halftoning and algorithm classifications are briefly introduced firstly. Then various error diffusion algorithms are proposed with its central parts of threshold and error filter; Structure-aware halftoning, dynamic global optimization. And Then the structure-aware error diffusion algorithm is analyzed from four respects: local frequency analysis, threshold modulation, diffusion filter modulation, and calibration. Further, the kernel size analysis and schematic of the structure-aware error diffusion algorithm are given, along with its pseudo code. At last, the comparison shows that the structure-aware error diffusion algorithm combines speed with high visual quality, and can be used in many practical applications that may require digital halftoning.