基于动态场景估计的自适应图像增强算法  

Adaptive Image Enhancement Algorithm Based on Dynamic Scene Estimation

作　　者：张超[1] 杨晶晶[1] 王盛[1] 陈更生[1] 

机构地区：[1]复旦大学专用集成电路与系统国家重点实验室,上海201203

出　　处：《计算机工程》2013年第5期34-41,共8页Computer Engineering

摘　　要：为快速、高效地增强因环境等因素导致质量退化的视频图像,提出一种基于动态场景估计的自适应图像增强算法。该算法采用拉普拉斯算子锐化图像的细节纹理,利用灰度统计的方法判据图像的场景,构建不同场景下的灰度映射函数,并根据动态场景估计的结果自适应地调整图像灰度的动态范围,以提升图像对比度。采用流水线处理技术和数据流行缓存技术完成硬件实现,执行效率高、占用资源少,可满足高清视频流的实时处理。实验结果表明,该算法在避免失真、保留原始语义的同时,能在各种场景下实现高质量的图像增强。In order to enhance efficiently a degraded video image in various scenes, a new adaptive image enhancement algorithm is proposed. The new algorithm uses Laplacian operator for extraction of image detail, constructs and uses different grey-level conversion strategy for images in different scenes based on dynamic seene estimation, and adaptively adjusts image＇s gray-level range to improve tile perlbrmance of image enhancement. By using line buffer and pipeline structure, the algorithm has a good timing performance and compact hardware implementation with minimum use of memory resource. Experimental result shows that the proposed algorithm has satisfying pertbrmance in each scene, while has a careful protection of image＇s original details.

关 键 词：动态场景估计 自适应图像增强 锐化 行缓存 流水线 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]