Search

A method and apparatus to extract a dense three-dimensional model of an observed scene or object from one or more sequences of images acquired with an imaging device such as a camera or camcorder, or clusters of cameras and/or camcorders. In some embodiments the method includes capturing an image sequence by a camera moving with a translateral motion, for example moving on a straight path with a fixed orientation. The images in each sequence are captured at regularly spaced sites on the straight path. In some embodiments, a device projects structured light on the observed scene. Multiple image sequences can be acquired, on different cameras on straight paths. The captured image sequences are input to a computer. The 3D structure of the scene viewed by each image sequence is computed. Dense reconstruction from EPIs is performed by interpolating the EPIs after having detecting straight segments on the EPIs. 3D models extracted from each image sequence are combined and merged to create a 3D model of the whole scene.

Shape recognition is the field of computer vision which addresses the problem of finding out whethera query shape lies or not in a shape database, up to a certain invariance. Most shape recognition methods simplysort shapes from the database along some (dis-)similarity measure to the query shape. Their main weakness is thedecision stage, which should aim at giving a clear-cut answer to the question: “do these two shapes look alike?”In this article, the proposed solution consists in bounding the number of false correspondences of the query shapeamong the database shapes, ensuring that the obtained matches are not likely to occur “by chance”. As an application,one can decide with a parameterless method whether any two digital images share some shapes or not