Showing papers on "Distance transform published in 1986"

Distance transformations in digital images

Abstract: A distance transformation converts a binary digital image, consisting of feature and non-feature pixels, into an image where all non-feature pixels have a value corresponding to the distance to the nearest feature pixel. Computing these distances is in principle a global operation. However, global operations are prohibitively costly. Therefore algorithms that consider only small neighborhoods, but still give a reasonable approximation of the Euclidean distance, are necessary. In the first part of this paper optimal distance transformations are developed. Local neighborhoods of sizes up to 7×7 pixels are used. First real-valued distance transformations are considered, and then the best integer approximations of them are computed. A new distance transformation is presented, that is easily computed and has a maximal error of about 2%. In the second part of the paper six different distance transformations, both old and new, are used for a few different applications. These applications show both that the choice of distance transformation is important, and that any of the six transformations may be the right choice.

Systolic Algorithms for Local Operations on Images.

Abstract: Les operateurs locaux utilises dans beaucoup de tâches de traitement d'image implique de remplacer chaque pixel dans une image par une valeur calculee a l'interieur d'un voisinage local de chaque pixel. Mise en œuvre d'un circuit VLSI

Method and apparatus for obtaining three dimensional distance information stereo vision

Abstract: A method of forming a three-dimensional stereo vision is disclosed in which, in order to prevent the erroneous detection of an object point in the conventional method using two image focusing lens systems, three image points of an object point formed by logically and/or physically selecting three image focusing lens systems and three image sensing surfaces from a multiplicity of image points on an image sensing surface by using the fact that a positional relation among three image points of the same object point is similar to the positional relation among the three image focusing lens systems, and the positional information of three selected image points on the image sensing surface is used for obtaining three-dimensional distance information of the object point.

Systolic Algorithnis for Local Operations on Images

Abstract: Local operators, used in many image processing tasks, involve replacing each pixel in an image with a value computed within a local neighborhood of that pixel. Computing such operators at the video rate requires a computing power which is not provided by conventional computers. Though computationally expensive, local operators are highly regular. Thus, a VLSI implementation appears particularly appropriate. This correspondence presents systolic algorithms for tasks such as connected component determination, distance transform, and relaxation, which are defined in terms of local operators.

A Skeletonizing Algorithm With Improved Isotropy

Abstract: An improved algorithm is presented which is capable of transforming thick objects in a discrete binaryimage into thinner representations called skeletons. The skeletal shapes produced are shown to be moreisotropic than those produced using other algorithms. The algorithm uses a non -iterative procedure based on the 4- distance ( "city block ") transform to produce connected reversible skeletons. The types and propertiesof 4- distance neighborhoods, which are used in skeletal pixel selection, are developed. Local- maxima are included in the skeleton, allowing reversibility using a reverse distance transform. Improved isotropy isachieved by defining pixels with certain types of neighborhoods to be interesting. It is shown that theseisotropy- improving pixels may be added to the skeletons produced by any 4- distance -based skeletonizingalgorithm that retains all local- maxima without affecting connectedness. I. IntroductionThick objects in a discrete binary image may be reduced to thinner representations called skeletons, whichare similar to stick figures. Most skeletonizing algorithms iteratively erode the contours in a binary imageuntil a thin skeleton remains. These pixel domain algorithms, which typically examine the neighborhood ofeach of the current contour pixels, may be divided into two groups: those which identify and remove"deletable" contour pixels, and those which identify skeletal pixels on the current contour and label themwith the current iteration number. The labels on skeletons of the latter type allow for partial or completereversibility (Section V). The execution times for these algorithms depend on the maximum thickness of theobjects being skeletonized.The execution times for skeletonizing algorithms that identify skeletal pixels by examining distanceneighborhoods, following a distance transform, may be made independent of object thickness. This isaccomplished by calculating the distance transform using one forward and one reverse raster scan [l].

A region extraction method using a distance function between images

Abstract: This paper discusses the distance function between images and its application, which is applicable to any monochrome binary image and monochrome gray-valued image with 2n × 2n pixels (n is a nonnegative integer). By denoting the original image as the 0th image, the first to the nth images are generated by successively partitioning the edges. Each image is composed of a finite number of subimages. Then the center of gravity is defined with the gray-value of the pixel as the weight. Using these notions, the distance between images is defined between two images. The rotational transformation is introduced for the subimages of the images at each stage, and the application of the distance function between images is discussed. An application example is the region extraction in the geographical information processing. The extraction of the contour (equal-height) line and the extraction of building region in the 1/25,000 map are discussed. The distance function between images proposed in this paper does not depend on the property of the original image, and is considered as being suited to the hierarchical processings.

Method for deriving 3-dimensional distance information

Abstract: @ A method of forming a three-dimensional stereo vision is disclosed in which, in order to prevent the erroneous detection of object point in the conventional method using two image focusing lens systems, three image points (c1, c2, c3) of an object point (c ) formed by logically and/or physically three image focusing lens systems (1, 2, 3) and three image sensing surfaces are selected from a multiplicity of image points on an image sensing surface by using the fact that a positional relation among three image points of the same object point is similar to the positional relation among the three image focusing lens systems, and the positional information of three selected image points on the image sensing surface is used for obtaining three-dimensional distance information of the object point (c).

Distance transformation and reconstructable skeleton of three‐dimensional digital binary images

Abstract: This paper considers the filled three-dimensional image, such as the one obtained by superposing several CT images, and discusses the distance transformation and the skeleton for the three-dimensional filled object. From the viewpoint that the internal structure of the three-dimensional object should easily be understood, there has been devised a distance transformation, where the distances from all points inside of the three-dimensional object to the surface are determined. There has not been presented, however, a method which is applicable to the general situation. From such a viewpoint, this paper introduces a new notion of fundamental neighbor set, and proposes a general parallel and serial algorithms for the distance transformation. As an example, the notion of the reconstruetable skeleton for the three-dimensional object is proposed, and the parallel and serial algorithms are presented, which can reconstruct accurately the original three-dimensional object from the skeleton. The reconstructable skeleton will be useful as a means of information compression for the three-dimensional image, which requires especially a large amount of data for the representation. The validity of the proposed algorithms is verified by computer simulation.