Orientation (computer vision)

About: Orientation (computer vision) is a research topic. Over the lifetime, 17196 publications have been published within this topic receiving 358181 citations.

Rectification of digital imagery

Abstract: Different methods can be applied to generate digital orthophotos. Three commonly used approaches are analyzed and compared in this paper. They can be applied to rectify both digitized aerial photographs and satellite scenes. These methods are polynomial, projective, and differential rectifications. The first two are defined by analytical transformations between image and orthophoto without considering the geometry and orientation of the camera. They are approximate solutions. The last one models the physical reality of the imaging process by means of the collinearity equations and corrects for relief displacements. All three methods were implemented on a workstation and were tested with digitized aerial photographs and video images. By overlaying GIS data over the digital orthophoto, the quality of the rectification is checked. To determine the planimetric accuracy of the results, the coordinates of targets were measured in a digital orthophoto and compared to known map coordinates.

Modeling and video projection for augmented virtual environments

Abstract: Systems and techniques to implement augmented virtual environments. In one implementation, the technique includes: generating a three dimensional (3D) model of an environment from range sensor information representing a height field for the environment, tracking orientation information of image sensors in the environment with respect to the 3D model in real-time, projecting real-time video from the image sensors onto the 3D model based on the tracked orientation information, and visualizing the 3D model with the projected real-time video. Generating the 3D model can involve parametric fitting of geometric primitives to the range sensor information. The technique can also include: identifying in real time a region in motion with respect to a background image in real-time video, the background image being a single distribution background dynamically modeled from a time average of the real-time video, and placing a surface that corresponds to the moving region in the 3D model.

Depth Estimation with Occlusion Modeling Using Light-Field Cameras

Abstract: Light-field cameras have become widely available in both consumer and industrial applications. However, most previous approaches do not model occlusions explicitly, and therefore fail to capture sharp object boundaries. A common assumption is that for a Lambertian scene, a pixel will exhibit photo-consistency, which means all viewpoints converge to a single point when focused to its depth. However, in the presence of occlusions this assumption fails to hold, making most current approaches unreliable precisely where accurate depth information is most important – at depth discontinuities. In this paper, an occlusion-aware depth estimation algorithm is developed; the method also enables identification of occlusion edges, which may be useful in other applications. It can be shown that although photo-consistency is not preserved for pixels at occlusions, it still holds in approximately half the viewpoints. Moreover, the line separating the two view regions (occluded object versus occluder) has the same orientation as that of the occlusion edge in the spatial domain. By ensuring photo-consistency in only the occluded view region, depth estimation can be improved. Occlusion predictions can also be computed and used for regularization. Experimental results show that our method outperforms current state-of-the-art light-field depth estimation algorithms, especially near occlusion boundaries.

Upright orientation of man-made objects

Abstract: Humans usually associate an upright orientation with objects, placing them in a way that they are most commonly seen in our surroundings. While it is an open challenge to recover the functionality of a shape from its geometry alone, this paper shows that it is often possible to infer its upright orientation by analyzing its geometry. Our key idea is to reduce the two-dimensional (spherical) orientation space to a small set of orientation candidates using functionality-related geometric properties of the object, and then determine the best orientation using an assessment function of several functional geometric attributes defined with respect to each candidate. Specifically we focus on obtaining the upright orientation for man-made objects that typically stand on some flat surface (ground, floor, table, etc.), which include the vast majority of objects in our everyday surroundings. For these types of models orientation candidates can be defined according to static equilibrium. For each candidate, we introduce a set of discriminative attributes linking shape to function. We learn an assessment function of these attributes from a training set using a combination of Random Forest classifier and Support Vector Machine classifier. Experiments demonstrate that our method generalizes well and achieves about 90% prediction accuracy for both a 10-fold cross-validation over the training set and a validation with an independent test set.

Video processing for composite images

Abstract: The image composition system includes framestores 30, 31 for receiving information from first and second picture sources. A processor 33 provides the composed image by using information from these sources. The processor is controlled by picture shape information made available from a third framestore 32. This shape information may be provided by a camera 26 receiving an image of a silhouette for example or the shape can be manually generated via a touch tablet 38. The instantaneous value of the shape controls the blending of the pictures such that portions of the picture can be taken from a scene and inserted without noticeable degredation. Manipulation of the position, orientation or size of the inserted picture portion for example can also be effected.