立体视觉舒适融合限预测模型的研究  

Research on the prediction model of vision comfortable fusion limits

作　　者：林丽媛[1] 侯春萍[1] 王凯[1] 

机构地区：[1]天津大学电子信息工程学院,天津300072

出　　处：《红外与激光工程》2014年第B12期231-237,共7页Infrared and Laser Engineering

基　　金：国家高技术研究发展计划(863计划)重大项目(2012AA03A301);国家自然科学基金重大研究计划重点项目(91320201);国家自然科学基金(61471262);教育部博士点基金(20110032110029;20130032110010)

摘　　要：针对导致立体显示中视觉舒适度问题的关键——立体视觉舒适融像问题,利用随机点立体图像(Random dot stereograms,RDS),对影响立体视觉舒适融像的因素进行了研究。通过改变RDS隐藏图形的大小,随机点的密度,随机点的大小和隐藏图形的形状,利用主观行为实验,分析这些因素对于立体视觉舒适融合限(Comfortable fusion limits,CFL)的影响,并提出关于CFL与这些因素的预测模型。实验结果表明,这些因素在交叉视差和非交叉视差情况下,对于立体视觉CFL的影响不同:非交叉视差下,被试者能够承受更大的隐藏图形,并且对于隐藏图形形状的CFL的感知与交叉视差相反;对由预测模型计算得出的CFL的预测值和实测值进行相关性检测,在交叉视差下,相关系数为0.998,非交叉视差下相关系数为0.977,可以认为该预测模型能够较为准确地预测CFL。Random dot stereograms were used to study the factors of the key problem that existed in the visual comfort of stereo display and was caused by stereo vision comfortable fusion. Subjective behavior experiments were done to analysis the factors to affect CFL（Comfortable fusion limits） of stereo vision by changing the size and shape of a hidden graph, the dots density and dot size in random dot stereograms.Then a prediction model of these factors to CFL was put forward. The results showed that these factors had different influences on CFL under crossed disparity and uncrossed disparity. Under uncrossed disparity, the subjects can stand bigger size of hidden graph, and the perception of fusion limit for the shape is opposite to the perception under crossed disparity. Correlation detection was measured between the prediction values calculated by the prediction model and the measured values. The correlation coefficient achieved 0.998 and 0.977, respectively, under crossed disparity and uncrossed disparity, which indicated that the prediction model could more accurately predict the CFL.

关 键 词：立体视觉 视觉舒适度 舒适融合限 随机点立体图 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]