Computer Graphics and Image Processing最新文献

Quadtrees offer a compact and hierarchical representation of an image. However, the location of an image segment on the coordinate grid and the grid size itself affect the quadtree representation, making comparison of image segments difficult. To alleviate this difficulty, we propose the concept of a normalized quadtree to represent a given image segment in a unique and optimal way over the class of all translations. An O(s²log₂s) algorithm is presented for obtaining normalized quadtrees, where s is the length of the grid.

A method for three-dimensional scene matching is presented in this paper. Both the geometric and structural information of the segmented features in two images are used for three-dimensional scene matching. The segmented features such as regions, edge segments, and vertices are initially labelled by using a symbol set. Then the structural relationships among these labels in each image are tabulated in a relational table. The consistent labels between two relational tables associated with two given images are searched using a relaxation labelling process. In this process, the matching line equation between the two images is used as a constraint function to remove the ambiguous labels from the two relational tables. This process is applied iteratively until two isomorphic relational tables are deduced. Since the labels in the two isomorphic tables are in one-to-one correspondence, the problem of matching the two images is reduced to a problem of matching regions to regions, edge segments to edge segments, and vertices to vertices. Again using the matching line equation as a geometric constraint function, the corresponding points in the two images may be searched in the corresponding edge segments. The three-dimensional object geometry is then computed using the matched corresponding points.

Two important features of the perception of plane curves are discontinuities in the curve and regions of high curvature. A method is presented, suitable for an open or closed curve, that given an ordered sequence of points will detect discontinuities in the perceived curve. The method is based on a local iterative process that attempts to obtain a set of cubic splines. The attempt fails when a sharp change occurs and this signals a discontinuity. Further iterations produce a smooth representation between the discontinuities. The given points may be irregularly spaced and no regular grid is superimposed on the data. The results obtained are orientation dependent and this appears to mimic our perception of curves.

In an image processing software system, when one wishes to perform a sequence of operations on an image, temporary files are typically used to pass the data from one program to the next. Pipelines, as implemented by the UNIX operating system, are an alternative to temporary files. Both approaches are compared in an image processing environment and it is shown that pipelines reduce the amount of processor time and clock time required.

A method of image segmentation was recently introduced based on defining links between pixels at adjacent levels of a “pyramid” of reduced-resolution versions of the image. This paper studies some of the problems that arise with linked-pyramid segmentation, and proposes a two-stage segmentation process that overcomes these problems.

An algorithm is presented that computes the intersection of two convex polygons in linear time. The algorithm is fundamentally different from the only known linear algorithms for this problem, due to Shamos and Hoey. These algorithms depend on a division of the plane into either angular sectors (Shamos) or parallel slabs (Hoey), and are mildly complex. Our algorithm searches for the intersection points of the polygons by adbancing a single pointer around each polygon, and is very easy to program.