A new formulation of the image partitioning problem is presented: construct a complete and stable description of an image, in terms of a specified descriptive language, that is simplest in the sense of being shortest. We show that a descriptive language limited to a low-order polynomial description of the intensity variation within each region and a chain-code-like description of the region boundaries yields intuitively satisfying partitions for a wide class of images. The advantage of this formulation is that it can be extended to deal with subsequent steps of the image-understanding problem (or to deal with other image attributes, such as texture) in a natural way by augmenting the descriptive language. Experiments performed on a variety of both real and synthetic images demonstrate the superior performance of this approach over partitioning techniques based on clustering vectors of local image attributes and standard edge-detection techniques. 1 Introduction The partitioning proble...

Our goal is to reconstruct both the shape and reflectance properties of surfaces from multiple images. We argue that an object-centered representation is most appropriate for this purpose because it naturally accommodates multiple sources of data, multiple images (including motion sequences of a rigid object), and self-occlusions. We then present a specific objectcentered reconstruction method and its implementation. The method begins with an initial estimate of surface shape provided, for example, by triangulating the result of conventional stereo. The surface shape and reflectance properties are then iteratively adjusted to minimize an objective function that combines information from multiple input images. The objective function is a weighted sum of stereo, shading, and smoothness components, where the weight varies over the surface. For example, the stereo component is weighted more strongly where the surface projects onto highly textured areas in the images, and less strongly othe...

A unified framework for 3-D shape reconstruction allows us to combine image-based and geometry-based information sources. The image information is akin to stereo and shape-from-shading, while the geometric information may be provided in the form of 3-D points, 3-D features or 2-D silhouettes. A formal integration framework is critical in recovering complicated surfaces because the information from a single source is often insufficient to provide a unique answer. Our approach to shape recovery is to deform a generic object-centered 3-D representation of the surface so as to minimize an objective function. This objective function is a weighted sum of the contributions of the various information sources. We describe these various terms individually, our weighting scheme, and our optimization method. Finally, we present results on anumber of difficult images of real scenes for which a single source of information would have proved insufficient.

Introduction A basic task for any visual system is to infer properties of the outside world given a stream of incoming images. The primary difficulty is that there are always an infinite number of combinations of world properties (e.g., surface shape, albedo, and illumination) that can produce the same image. Thus, one cannot simply deduce these properties from an image. Instead, one must choose a single combination according to some guiding principle. We hypothesize that an important guiding principle in vision is simplicity of description, or, more formally, "minimal-length encoding" (Chaitin 1966, Minsky 1962, Rissanen 1978, Solomonoff 1964). According to this principle, prior information about the world and image sensor is incorporated in the language used to describe the world and sensor, and the inference process is to find the simplest (i.e., shortest) description in the language that exactly reproduces the given images. The basic motivation behind this inference process is th