Abstract: A system and process for computing a 3D reconstruction of a scene from multiple images thereof, which is based on a color segmentation-based approach, is presented. First, each image is independently segmented. Second, an initial disparity space distribution (DSD) is computed for each segment, using the assumption that all pixels within a segment have the same disparity. Next, each segment's DSD is refined using neighboring segments and its projection into other images. The assumption that each segment has a single disparity is then relaxed during a disparity smoothing stage. The result is a disparity map for each image, which in turn can be used to compute a per pixel depth map if the reconstruction application calls for it.

Abstract: The present invention is embodied in a system and method for statistically comparing a first set of digital data to at least a second set of digital data and matching the first set of digital data to appropriately corresponding portions of the second set of digital data. The first or the second set of digital data can be transformed during statistical analysis to enhance statistical analysis of the digital data.

Abstract: A system and process for generating, and then rendering and displaying, an interactive viewpoint video in which a user can watch a dynamic scene while manipulating (freezing, slowing down, or reversing) time and changing the viewpoint at will. In general, the interactive viewpoint video is generated using a small number of cameras to capture multiple video streams. A multi-view 3D reconstruction and matting technique is employed to create a layered representation of the video frames that enables both efficient compression and interactive playback of the captured dynamic scene, while at the same time allowing for real-time rendering.

Abstract: A system and process for generating, and then rendering and displaying, an interactive viewpoint video in which a user can watch a dynamic scene while manipulating (freezing, slowing down, or reversing) time and changing the viewpoint at will. In general, the interactive viewpoint video is generated using a small number of cameras to capture multiple video streams. A multi-view 3D reconstruction and matting technique is employed to create a layered representation of the video frames that enables both efficient compression and interactive playback of the captured dynamic scene, while at the same time allowing for real-time rendering.

Abstract: An “Oblique Image Stitcher” provides a technique for constructing a photorealistic oblique view from a set of input images representing a series of partially overlapping views of a scene. The Oblique Image Stitcher first projects each input image onto a geometric proxy of the scene and renders the images from a desired viewpoint. Once the images have been projected onto the geometric proxy, the rendered images are evaluated to identify optimum seams along which the various images are to be blended. Once the optimum seams are selected, the images are remapped relative to those seams by leaving the mapping unchanged at the seams and interpolating a smooth mapping between the seams. The remapped images are then composited to construct the final mosaiced oblique view of the scene. The result is a mosaic image constructed by warping the input images in a photorealistic manner which agrees at seams between images.

Abstract: An “Image Denoiser” provides a probabilistic process for denoising color images by segmenting an input image into regions, estimating statistics within each region, and then estimating a clean (or denoised) image using a probabilistic model of image formation. In one embodiment, estimated blur between each region is used to reduce artificial sharpening of region boundaries resulting from denoising the input image. In further embodiments, the estimated blur is used for additional purposes, including sharpening edges between one or more regions, and selectively blurring or sharpening one or more specific regions of the image (i.e., “selective focus”) while maintaining the original blurring between the various regions.

Abstract: A feature symbol triplets object instance recognizer and method for recognizing specific objects in a query image. Generally, the recognizer and method find repeatable features in the image, and match the repeatable features between a query image and a set of training images. More specifically, the recognizer and method finds features in the query image and then groups all possible combinations of three features in to feature triplets. Small regions or “patches” in the query image, and an affine transformation is applied to the patches to identify any similarity between patches in a query image and training images. The affine transformation is computed using position of neighboring features in each feature triplet. Next, all similar patches are found, and then pairs of images are aligned to determine if the patches agree in the position of the object. If they do, then it is said that object is found and identified.

Abstract: A Bayesian two-color image demosaicer and method for processing a digital color image to demosaic the image in such a way as to reduce image artifacts. The method and system are an improvement on and an enhancement to previous demosaicing techniques. A preliminary demosaicing pass is performed on the image to assign each pixel a fully specified RGB triple color value. The final color value of pixel in the processed image is restricted to be a linear combination of two colors. Fully-specified RGB triple color values for each pixel in an image used to find two clusters represented favored two colors. The amount of contribution from these favored two colors on the final color value then is determined. The method and system also can process multiple images to improve the demosaicing results. When using multiple images, sampling can be performed at a finer resolution, known as super resolution.

Abstract: A system and process for generating High Dynamic Range (HDR) video is presented which involves first capturing a video image sequence while varying the exposure so as to alternate between frames having a shorter and longer exposure. The exposure for each frame is set prior to it being captured as a function of the pixel brightness distribution in preceding frames. Next, for each frame of the video, the corresponding pixels between the frame under consideration and both preceding and subsequent frames are identified. For each corresponding pixel set, at least one pixel is identified as representing a trustworthy pixel. The pixel color information associated with the trustworthy pixels is then employed to compute a radiance value for each pixel set to form a radiance map. A tone mapping procedure can then be performed to convert the radiance map into an 8-bit representation of the HDR frame.

Abstract: A system and process for rendering and displaying an interactive viewpoint video is presented in which a user can watch a dynamic scene while manipulating (freezing, slowing down, or reversing) time and changing the viewpoint at will. The ability to interactively control viewpoint while watching a video is an exciting new application for image-based rendering. Because any intermediate view can be synthesized at any time, with the potential for space-time manipulation, this type of video has been dubbed interactive viewpoint video.

Abstract: Over the past few years there has been a dramatic proliferation of digital cameras, and it has become increasingly easy to share large numbers of photographs with many other people. These trends have contributed to the availability of large databases of photographs. Effectively organizing, browsing, and visualizing such .seas. of images, as well as finding a particular image, can be difficult tasks. In this paper, we demonstrate that knowledge of where images were taken and where they were pointed makes it possible to visualize large sets of photographs in powerful, intuitive new ways. We present and evaluate a set of novel tools that use location and orientation information, derived semi-automatically using structure from motion, to enhance the experience of exploring such large collections of images.

Abstract: A system and process for reconstructing optimal texture maps from multiple views of a scene is described. In essence, this reconstruction is based on the optimal synthesis of textures from multiple sources. This is generally accomplished using basic image processing theory to derive the correct weights for blending the multiple views. Namely, the steps of reconstructing, warping, prefiltering, and resampling are followed in order to warp reference textures to a desired location, and to compute spatially-variant weights for optimal blending. These weights take into consideration the anisotropy in the texture projection and changes in sampling frequency due to foreshortening. The weights are combined and the computation of the optimal texture is treated as a restoration problem, which involves solving a linear system of equations.

Abstract: A system and process for generating a two-layer, 3D representation of a digital or digitized image from the image and a pixel disparity map of the image is presented. The two layer representation includes a main layer having pixels exhibiting background colors and background disparities associated with correspondingly located pixels of depth discontinuity areas in the image, as well as pixels exhibiting colors and disparities associated with correspondingly located pixels of the image not found in these depth discontinuity areas. The other layer is a boundary layer made up of pixels exhibiting foreground colors, foreground disparities and alpha values associated with the correspondingly located pixels of the depth discontinuity areas. The depth discontinuity areas correspond to prescribed sized areas surrounding depth discontinuities found in the image using a disparity map thereof.

Abstract: The claimed subject matter can provide an architecture that facilitates producing a single image that can visualize a scene too large to depict from any single perspective view. These images can be stitched together on a 2-dimensional picture surface to form a multi-perspective image of the entire extent of the scene depicted by the input images. Where the respective images overlap, an objective function can be employed to determine where to make a transition between one image and the next. The objective function can employ a data cost and a seam cost to make this determination.

Abstract: A “Keyframe Stitcher” provides an efficient technique for building mosaic panoramic images by registering or aligning video frames to construct a mosaic panoramic representation. Matching of image pairs is performed by extracting feature points from every image frame and matching those points between image pairs. Further, the Keyframe Stitcher preserves accuracy of image stitching when matching image pairs by utilizing ordering information inherent in the video. The cost of searching for matches between image frames is reduced by identifying “keyframes” based on computed image-to-image overlap. Keyframes are then matched to all other keyframes, but intermediate image frames are only matched to temporally neighboring keyframes and neighboring intermediate frames to construct a “match structure.” Image orientations are then estimated from this match structure and used to construct the mosaic.

Abstract: A “Panoramic Viewfinder” provides an intuitive interactive viewfinder display which operates on a digital camera display screen. This interactive viewfinder provides real-time assistance in capturing images for constructing panoramic image mosaics. The Panoramic Viewfinder “brushes” a panorama from images captured in any order, while providing visual feedback to the user for ensuring that desired scene elements will appear in the final panorama. This visual feedback presents real-time stitched previews of the panorama while capturing images.

Abstract: A system and process for rendering and displaying an interactive viewpoint video is presented in which a user can watch a dynamic scene while manipulating (freezing, slowing down, or reversing) time and changing the viewpoint at will. The ability to interactively control viewpoint while watching a video is an exciting new application for image-based rendering. Because any intermediate view can be synthesized at any time, with the potential for space-time manipulation, this type of video has been dubbed interactive viewpoint video.

Abstract: A system and process for generating a high dynamic range (HDR) image from a bracketed image sequence, even in the presence of scene or camera motion, is presented. This is accomplished by first selecting one of the images as a reference image. Then, each non-reference image is registered with another one of the images, including the reference image, which exhibits an exposure that is both closer to that of the reference image than the image under consideration and closest among the other images to the exposure of the image under consideration, to generate a flow field. The flow fields generated for the non-reference images not already registered with the reference image are concatenated to register each of them with the reference image. Each non-reference image is then warped using its associated flow field. The reference image and the warped images are combined to create a radiance map representing the HDR image.

Abstract: An automatic digital image grouping system and method for automatically generating groupings of related images based on criteria that includes image metadata and spatial information. The system and method takes an unordered and unorganized set of digital images and organizes and groups related images into image subsets. The criteria for defining an image subset varies and can be customized depending on the needs of the user. Metadata (such as EXIF tags) already embedded inside the images is used to extract likely image subsets. This metadata may include the temporal proximity of images, focal length, color overlap, and geographical location. The first component of the automatic image grouping system and method is a subset image stage that analyzes the metadata and generates potential image subsets containing related images. The second component is an overlap detection stage, where potential image subset is analyzed and verified by examining pixels of the related images.

Abstract: A panoramic high-dynamic range (HDR) image method and system of combining multiple images having different exposures and at least partial spatial overlap wherein each of the images may have scene motion, camera motion, or both. The major part of the panoramic HDR image method and system is a two-pass optimization-based approach that first defines the position of the objects in a scene and then fills in the dynamic range when possible and consistent. Data costs are created to encourage radiance values that are both consistent with object placement (defined by the first pass) and of a higher signal-to-noise ratio. Seam costs are used to ensure that transitions occur in regions of consistent radiances. The result is a high-quality panoramic HDR image having the full available spatial extent of the scene along with the full available exposure range.