权益分类	功能权益	普通用户	{{item.name}}会员
{{category.name}}	{{benefitItem.name}}

Stereoscopic Image Synthesis Based On Camera and Eye Calibrations

基于相机和眼睛校准的立体图像合成

基本信息

批准号：
12680357
负责人：
ASADA Naoki
金额：
$ 2.37万
依托单位：
Hiroshima City University
依托单位国家：
日本
项目类别：
Grant-in-Aid for Scientific Research (C)
财政年份：
2000
资助国家：
日本
起止时间：
2000 至 2002
项目状态：
已结题

来源：
https://kaken.nii.ac.jp/grant/KAKENHI-PROJECT-12680357/
关键词：
Camera model Camera calibration Stereopsis Image synthesis Computer vision Computer graphics Virtual reality Algorithm ぼけ画像距離計測

项目摘要

We have investigated the theory and application of stereoscopic image synthesis based on camera and eye calibrations.1. Evaluation of the unified camera model and its calibration procedure :We have discussed the optical properties of a zoom lens system, and proposed a unified camera model that is characterized by the effective focal length and lens diameter. Using the camera model, we have proved that the depth information is recovered from the image blur at any setting of camera parameters. To evaluate the precision of camera model and calibration results, we have examined the accuracy of estimated depth and shown an application of the depth information to image synthesis.2. Evaluation of image generation algorithm :We have implemented two image generation algorithms called multiple Z-buffer algorithm and multiple pinhole one that are developed based on the unified camera model. Experimental evaluation on the accuracy and performance of them have shown that both algorithms are useful for the high quality image generation with lens effects.3. Development of eye calibration procedure for stereoscopic image synthesis :We have developed an eye calibration procedure to determine the individual eye parameters and shown some experimental results including a stereoscopic image synthesized from a stereo pair of real images with virtual objects.4. Application of eye calibration to cataract operation simulator :We have applied the eye calibration method to the cataract operation simulator in which the stereoscopic images are generated as a microscopic view. Experimental results have demonstrated the scale variability of the eye calibration method.

研究了基于摄像机和人眼标定的立体图像合成的理论和应用。对统一相机模型及其标定过程的评价：讨论了变焦镜头系统的光学特性，提出了以有效焦距和镜头直径为特征的统一相机模型。利用摄像机模型，我们证明了在摄像机参数的任意设置下，都能从图像模糊中恢复出深度信息。为了评价摄像机模型和标定结果的精度，我们检验了深度估计的准确性，并展示了深度信息在图像合成中的应用。图像生成算法的评价：我们实现了基于统一摄像机模型的多Z-Buffer算法和多针孔算法。对这两种算法的精度和性能的实验评估表明，这两种算法对于生成具有镜头效果的高质量图像是有用的。开发了用于立体图像合成的人眼定标程序：我们开发了一种确定单个人眼参数的人眼定标程序，并给出了一些实验结果，包括由一对含有虚拟物体的立体图像合成的立体图像。人眼标定在白内障手术模拟器中的应用：我们将人眼标定方法应用到白内障手术模拟器中，将立体图像生成为显微图像。实验结果证明了人眼定标方法的尺度可变性。