Integration Methods for Reliable Information Extraction from Multiple Color Images

我们研究了系统地改变相机参数、光圈和焦距所拍摄的多幅图像的集成方法。研究报告描述了多虹膜和多焦点图像的分析，并展示了集成方法，以提取可靠的场景信息，如色度，亮度，边缘，深度等。1.多虹膜彩色图像的分析我们提出了一种方法，以获得可靠的颜色信息，即色度和亮度，从不同孔径大小的多个彩色图像。我们还展示了多虹膜彩色图像的其他用途：光源色度估计和相对深度的推导。2.多聚焦图像的分析我们对多聚焦图像沿聚焦轴沿着进行分析，得到两种表示综合信息的图像;一个是其零交叉对应于场景边缘的累积散焦图像，3.多聚焦图像中遮挡边缘特性的分析提出了一种基于光通量的模糊模型--逆投影模糊模型，用于分析模糊遮挡边缘的光度特性。使用该模型，我们表明：（1）的光学图像的遮挡边缘与表面边缘的较近的物体重合，如果较远的物体的辐射是均匀的周围的遮挡边缘，（2）复杂的局部强度极值，即谷和峰，被引入到一个遮挡边缘，如果表面边缘位于较远的物体周围的遮挡边缘。

We have investigated the integration methods of multiple images taken by systematically varying camera parameters, iris and focus. The research report describes the analyzes of multiiris and multi-focus images and shows integration methods to extract reliable scene information such as chromaticity, brightness, edge, depth, and so on.1.Analysis of Multi-Iris Color ImagesWe proposed a method to obtain the reliable color information, i.e. chromaticity and brightness, from multiple color images taken with different aperture sizes. We also demonstrate other utilities of the multi-iris color images : estimation of light source chromaticity and derivation of relative depth.2.Analysis of Multi-Focus ImagesWe analyzed multi-focus images along the focus axis to yield two types of images representing integrated information ; one is an accumulated defocus image whose zero-crossings correspond to edges of the scene, and the other a spatio-focal image with which the depth of the edges can be estimated.3.Analysis of Occluding Edge Properties in Multi-Focus ImagesWe proposed an optical flux based blurring model named the reversed projection blurring model to analyze photometric properties of blurred occluding edges. Using the proposed model, we showed that (1) the optical image of an occluding edge coincides with that of a surface edge on the nearer object, if the radiance of the farther object is uniform around the occluding edge, (2) complex local intensity extrema, i.e. valleys and peaks, are introduced into an occluding edge, if a surface edge is located on the farther object around the occluding edge.