基于运动和散焦线索的深度提取研究  

Depth Extraction based on Motion and Defocusing Cue

作　　者：牛连丁[1] 孙华东[2] 陈铭[2] 

机构地区：[1]哈尔滨商业大学网络与教育技术中心,哈尔滨150028 [2]哈尔滨商业大学计算机与信息工程学院,哈尔滨150028

出　　处：《智能计算机与应用》2015年第2期18-20,24,共4页Intelligent Computer and Applications

基　　金：黑龙江省自然科学基金(F201114);哈尔滨市科技创新人才研究专项资金(2012RFQXG090)

摘　　要：随着立体图像在社会各个领域的发展与普及,三维资源的需求逐步提升,而将二维图像转换为三维图像是解决三维资源稀缺的重要手段。二维图像三维化技术的关键在于深度信息的获取。分析了两种重要的深度提取线索,运动线索和散焦模糊线索。为了弥补这两个线索在深度提取的局限性,提出了两种线索的深度融合策略。实验结果表明,这两种线索采用分段线性融合的方法能提高生成深度图的质量。With the development and popularity of stereo image in field of society,the demand of stereo resource gradually upgrades. 2D to 3D image conversion is an important methods to solve the problem of stereo resource scarcity. As a new interactive way,3D visualization technology eliminates the disadvantages of plane communication interface. Depth extraction is the key for three dimensional visualization technique. The two important cues,motion cue and defocusing cue are analyzed. In order to compensate for the limitations of the two cues,the fusion strategy of the two cues is presented here. Experimental results show that piecewise linear combination of the two cues can improve the quality of depth map.

关 键 词：深度提取 运动线索 散焦线索 深度融合策略 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]