No abstract

Abstract-Computervision systems attempt to recover useful information about the three-dimensional world from huge image arrays of sensed values. Since direct interpretation of large amounts of raw data by computer is difficult, it is often convenient to partition (segment) image arrays into low-level entities (groups of pixels with similar properties) that can be compared to higher-level entities derived from representations of world knowledge. Solving the segmentation problem requires a mechanism for partitioning the image array into low-level entities based on a model of the underlying image structure. Using a piecewise-smooth surface model for image data that possesses surface coherence properties, we have developed an algorithm that simultaneously segments a large class of images into regions of arbitrary shape and approximates image data with bivariate functions so that it is possible to compute a complete, noiseless image reconstruction based on the extracted functions and regions. Surface curvature sign labeling provides an initial coarse image segmentation, which is refined by an iterative region growing method based on variable-order surface fitting. Experimental results show the algorithm's performance on six range images and three intensity images.Index Terms--Image segmentation, range images, surface fitting.I. INTR~DUCTI~N C OMPUTER vision systems attempt to recover useful information about the three-dimensional (3-D) world from huge image arrays of sensed values. Since direct interpretation of large amounts of raw data by computer is difficult, it is often convenient to partition (segment) image arrays into low-level entities (groups of pixels with particular properties) that can be compared to higher-level entities derived from representations of world knowledge. Solving the segmentation problem requires a mechanism for partitioning the image array into useful entities based on a model of the underlying image structure.In most easily interpretable images, almost all pixel values are statistically and geometrically correlated with neighboring pixel values. This pixel-to-pixel correlation, or spatial coherence, in images arises from the spatial coherence of the physical surfaces being imaged. In range images, where each sensed value measures the distance to physical surfaces from a known reference surface, the pixel values collectively exhibit the same spatial coher- ence properties as the actual physical surfaces they represent. This has motivated us to explore the possibilities of a surface-based image segmentation algorithm that uses the spatial coherence (surface coherence) of the data to organize pixels into meaningful groups for later visual processes. Many computer vision algorithms are based on inflexible, unnecessarily restricting assumptions about the world and the underlying structure of the sensed image data. The following assumptions are common: 1) all physical objects of interest are polyhedral, quadric, swept (as in generalized cylinders), convex, or combinations thereof; 2) all physi...

This paper describes a general purpose, representation independent method for the accurate and computationally efficient registration of three-dimensional (3-D) shapes including free-form curves and surfaces. The method handles the full six-degrees of freedom and is based on the Iterative Closest Point (ICP) algorithm, which requires only a procedure to find the closest point on a geometric entity to a given point. The ICP algorithm always converges monotonically to the nearest local mmimum of a mean-square distance metric, and experience shows that the rate ofconvergence is rapid during the first few iterations. Therefore, given an adequate set of initial rotations and translations for a particular class of objects with a certain level of "shape complexity" ,one can globally minimize the mean-square distance metric over all six degrees of freedom by testing each initial registration. For example, a given "model" shape and a sensed "data" shape that represents a major portion of the model shape can be registered in minutes by testing one initial translation and a relatively small set of rotations to allow for the given level of model complexity. One important application of this method is to register sensed data from unfixtured rigid objects with an ideal geometric model prior to shape inspection. The described method is also useful for deciding fundamental issues such as the congruence ( shape equivalence) of different geometric representations as well as for estimating the motion between point sets where the correspondences are not known. Experimental results show the capabilities of the registration algorithm on point sets, curves, and surfaces.

A general-purpose computer vision system must be capable of recognizing three-dimensional (3-D) objects. This paper proposes a precise definition of the 3-D object recognition problem, discusses basic concepts associated with this problem, and reviews the relevant literature. Because range images (or depth maps) are often used as sensor input instead of intensity images, techniques for obtaining, processing, and characterizing range data are also surveyed.