Abstract: A contour approximation apparatus for representing a contour image of an object comprises a polygonal approximation section for determining a number of vertices on the contour image and fitting the contour image with a plurality of line segments to provide a polygonal approximation of the contour image, a sampling circuit for providing N sample points for each of the line segments, an error detector for calculating an error for each of the N sample points on each of the line segments to produce a set of errors for each of the line segments, a discrete sine transform and quantization block for transforming each set of errors into a set of discrete sine transform coefficients, and for converting the set of discrete sine transform coefficients into a set of quantized transform coefficients.

Abstract: The feature points of the input video signal are detected by canny edge detecting method, the coordinates of the feature points detected are chain coded, the length of the region, including the feature points, of which edge intensity is the specified value or more in the edge intensity maximum change direction at the feature point coordinates obtained when the feature point is detected, are code, and the chain-coded data and the region length coded data are then multiplexed. Thereby, the amount of bits required for coding of the coordinates information can be reduced, as a result the total amount of codes can be reduced.

Abstract: There is provided an improved scan interleaving method suitable for multipurpose encoding of binary image data, and for selectively encoding a part or the whole of object image data according to presence or absence of exceptional sampling data (ESD). In the present invention, when the pixel data of base image data and the corresponding pixel data of the object image data to be scan-interleaved are the same (in the case of predictive sampling data), no encoding is performed on the pixels of the object image data which have no ESD, thus simplifying the algorithm and system implementation. As a result, the overall encoding performance can be improved.

Abstract: A method and apparatus is disclosed for converting raster images into a vector format by identifying and converting the borders of image features into a mathematical format referred to as string sequences. To enable the string sequencing of color images, the present invention identifies and resolves contrast conflicts in the image features in order to avoid misperceiving the vectorized image when converted back into a raster format. A contrast conflict occurs when there is a “contrast tie” between overlapping features of the image. The feature that would normally be perceived as the dominant feature breaks the contrast tie so that when the vector image is reconstructed, the dominate feature appears in the foreground of the image while the recessive feature appears in the background.

Abstract: Noise-corrupted boundary information in tomographic images is restored using a data structure, called a “boundary envelope”. Both the restored boundary information and a measure of the precision of the restoration are described by the process. This data structure is used to describe the restored information in both qualitative and quantitative terms. The results of the boundary restoration process are described qualitatively by graphically displaying the boundary envelope together with the noisy CT image in such a way that allows a human observer to (a) see more details of the boundary than are visible in the CT image alone, and (b) to visually apprehend the precision with which the boundary has been restored. A boundary envelope describes a region in the image within which the boundary of a given noisy object is guaranteed to lie. The boundary envelope describes a set of limits on the object boundary, and limits on those properties of a noisy object that can be computed from boundary information.

Abstract: An object of the present invention is to provide an image cutout method capable of moving a nozzle on a photographic image along a contour of a subject and binarizing image data in the nozzle to obtain the contour line of the subject, and cutting out the image along the obtained contour line, and as a result, allowing efficient cutout without putting a load on an operator. To achieve the object, when the nozzle is moved, a direction for moving the nozzle is determined based on a partial contour line in the obtained present nozzle. Thus, it is possible to move the nozzle along the contour of the subject without operator's operation of a pointing device and the like.

Abstract: A method for compressing a mesh having a plurality of vertices, each vertex characterized by a degree equal to the number of edges incident thereon, including arranging substantially all of the vertices in a consecutive order, generating a topology list including the degrees of the vertices in the consecutive order, and providing a coded stream of signals including the topology list.

Abstract: It is an object of the present invention to provide a method for automatically segmenting image objects by applying contour tracking to image objects with segmentation information of previous image object, and according to an aspect of the present invention, there is provided a method for segmenting an image object of a moving picture including the steps of: extracting and storing a contour of a previous image object of a previous scene by using a predetermined segmentation information of the previous image object; calculating a motion information by extracting a contour of the previous image object at a current scene; extracting a contour of a current image object by using the motion information and a space domain information of previous and current scenes; and extracting an image object area of the current image object at the current scene by using the contour of the current image object.

Abstract: A vertex-based shape coding method and an inter-frame vertex predicting method are provided.

Abstract: A cost effective continuously adaptive digital video system and method for compressing color video data for moving images. The method describes capturing an analog video frame and digitizing the image into a preferred source input format for compression using a combination of unique lossy and lossless digital compression techniques including subband coding, wavelet transforms, motion detection, run length coding and variable length coding. The system includes encoder and decoder (CODEC) sections for compression and decompression of visual images to provide high compression with good to excellent video quality. The compressed video data provides a base video layer and additional layers of video data that are multiplexed with compressed digital audio to provide a data stream that can be packetized for distribution over inter or intranets, including wireless networks over local or wide areas.

Abstract: An object of this invention is to provide a coding method and a coding apparatus for efficiently coding both a pixel value signal and a shape signal which are included in an image signal to be coded, wherein a prediction changing unit outputs a reference pixel value changing signal and a reference shape signal to a switching circuit which selects a reference pixel value signal and to a switching circuit which selects a reference shape signal, respectively, so as to control the switching circuit in a way that appropriate reference signals are to be selected.

Abstract: In an apparatus for adaptively coding an image signal including a contour signal and a texture signal having object pixels and background pixels thereof, wherein the contour signal discriminates between the object pixels and the background pixels and each of the background pixels and the object pixels is represented by luminance data and chrominance data, a padding circuit replaces each background pixel value with a pixel value derived from object pixel values by using a predetermined padding method to thereby generate a padded texture signal. And then, a luminance data extracting circuit extracts the luminance data from the padded texture signal to provide a modified texture signal, each pixel of the modified texture signal having the luminance data.

Abstract: A method encodes a contour image of an object expressed in a video signal. First, a centroid for the contour of the object is detected by averaging pixel positions on the contour. A set of primary vertices on the contour are determined based on the centroid. Subsequently, a set of secondary vertices on the contour are determined based on the set of primary vertices. The set of primary vertices, the set of secondary vertices and the centroid are encoded to provide a digitally coded contour signal.

Abstract: A method for encoding an input image including more than an object is provided, wherein if number of contours in an image frame is less than a predetermined number, all the absolute addresses of the initial vertices of the contours are encoded. However, if the number of the contours are greater than or equal to the predetermined number, the contours are divided into a plurality of groups, wherein if a position of an initial vertex in a group is known, relative displacements of all the other vertices in the group can be represented by dynamic ranges determined by the number of the contours. Thereafter, absolute position information of a first initial vertex in each group is encoded using an escape code which is followed by the horizontal and the vertical absolute addresses of the initial vertex. All the remaining initial vertices are encoded based on their displacements from the previously encoded initial vertex and the dynamic ranges.

Abstract: A contour decoding apparatus decodes a bitstream of contour information which was encoded using a chain code. A demultiplexing unit separates the bitstream of the contour information into start contour pixel information and chain code information. A start contour pixel decoder decodes start contour pixel information separated by the demultiplexing unit and outputs horizontal and vertical coordinate values of the start pixel.

Abstract: In a coding system, an input picture is split into regions with adaptive configurations to a reference picture, a selected region with precedence data is detected of its configuration and of its relative motion to a corresponding region of the reference picture, a prediction region is generated from the corresponding region of the reference picture and the relative motion, a difference region is determined between the selected region and the prediction region, and the precedence data, detected configuration, relative motion and difference ergion are coded. In a decoding system, a decoded prediction region is generated from a decoded relative motion and a corresponding region of a reference picture selected by a decoded precedence data, and a decoded difference region is added to the decoded prediction region to obtain a decoded picture region, which is configuration-corrected by a decoded configuratrion data, before synthesizing a plurality of decoded picture regions to obtain a synthesized picture region.

Abstract: A method for encoding input contour image including a current contour based on a previous contour, the contour encoded previously, is provided. In this method a predicted contour is generated from motion estimating and compensating the previous contour based on the current contour. By indexing the contour pixels on the predicted contour and encoding the indices of the end-points of matched segments, parts where the predicted and the current contours overlap, encoded major vertex information is provided. Unmatched segments, parts on the current contour which are not included on any of the matching segments, are polygonal approximated to find minor vertices and relative location of the minor vertices are encoded to provide encoded minor vertex information. When encoding the minor vertices, modes for encoding the number of minor vertices are selected to minimize the number of bits for encoding the number of minor vertices.

Abstract: A method for coding multiple layer, vertex-based shape representation and bit stream scalable shape includes selecting vertices in a hierarchical scheme; defining multiple layers of vertices; arranging M layers of vertices in order, from the most salient layer to the least salient layer; placing vertices in their natural order along their image contour within each layer; storing vertices from each layer separately; and recording vertex order information for each vertex in each layer of vertices. Four variations of the method for multiple layer vertex representation and coding are disclosed where the correct order information is coded for correct reconstruction of the shape information.

Abstract: A method for compressing the bit data of an alpha mask image frame composed of an array of pixels having bit data indicative of a level of brightness includes selecting a modifying function identifier which represents a mathematical transformation, applying a threshold function to the alpha mask image frame to generate a binary image frame, and encoding the binary image frame and the modifying function identifier for transmission. The method may include receiving an encoded signal containing the binary image frame and the modifying function identifier, decoding the encoded signal to extract the binary image frame and the modifying function identifier, selecting the mathematical transformation represented by the modifying function identifier, and then transforming the bit data of at least one pixel in accordance with the mathematical transformation.

Abstract: A chain code process for encoding or compressing contours of masks, objects, sprites, and other graphical features, limits the number of chain codes and defines them relative to the alignment of adjacent pairs of pixels. The chain code direction for a pair of contour pixels is defined relative to the alignment direction of an immediately adjacent pair of pixels. A three-point chain code of this invention includes chain codes that correspond to contour pixel alignments in a forward direction, a leftward direction, and a rightward direction. The forward direction indicates that the alignment direction for a pair of contour pixels is the same as the alignment direction of an immediately adjacent pair of pixels. The rightward and leftward alignment directions indicate that the alignment direction for a pair of contour pixels deviate to the right and left, respectively, from the forward direction.

Abstract: A method is provided for reducing the number of data bits required for storage or transmission of the data representing the alpha masks of objects in a single image or frame of a video sequence by jointly encoding overlapping boundaries between two adjacent objects within the image and by avoiding redundant encoding and transmission of an alpha mask which can be reconstructed from other encoded alpha mask data.

Abstract: In a picture encoding and decoding method and apparatus, an object contour can be chain coded precisely even in the case where noise exists in the vicinity of object contour, and the object contour branched off can be coded with small quantity of codes. In the case where the branched part occurs in a feature point chain, by multiplexing branch code D1 to chain codes S23 and S22 and expressing as a tree structured feature point chain including the branch part of the object contour, an increase in number of chains can be controlled and the efficient chain coding can be conducted. Moreover, since the terminal branch chains having no branch point in the chain become the objects of thresholding, only noise elements can be effectively eliminated without cutting off the feature point chain showing the object contours.

Abstract: An apparatus polygonally approximates a contour of an object expressed in a digital video signal. First of all, the apparatus determines a pair of vertices on the contour. Then, a line segment connecting the pair of vertices is generated and widened to thereby produce a band segment. And, a contour segment corresponding to the line segment is detected and, subsequently, the band segment and the contour segment are matched. The band segment is widened to determine a new vertex located between the pair of vertices on the contour until the contour segment is completely covered by the widened band segment. During the band segment widening process, at least one contour pixel on the contour segment, wherein the detected contour pixels, if there are more than one, are last pixels being covered by the widened band segment, are determined and, among the detected contour pixels, a contour pixel, which is closest to a center of a straight line joining the pair of vertices, is determined as the new vertex.

Abstract: An image data coding system comprises a screen-area determining means for determining the area of a screen reproduced on the basis of an input image data; a code-amount assigning control means for controlling regions to which data on the screen are to be assigned and the code amount assigned to each of the regions, on the basis of the results determined on the area of the reproduced screen; and a coding means for coding the image data signal inputted in accordance with the code amount assigned to each of the regions. In an image data coding and/or decoding system, a coding system performs the two-dimensional orthogonal transform of picture signals of all the pixels with respect to inside blocks and of only picture signals of pixels contained in a content with respect to edge blocks, in accordance with a map signal indicative of the position and shape of the content, and it codes the map signal.

Abstract: Baseline-based contour coding and decoding methods are provided. The coding method comprises the steps of determining a baseline of a still image or moving image object and a predetermined sampling interval, sampling the contour of the object according to the baseline and the sampling interval and extracting contour sample data, extracting utmost contour sample data of the object from the contour sample data, and calculating a difference between the utmost contour sample data and the contour sample data and extracting error sample data. Using a characteristics of contour data having a gentle change, the still image contour coding using transformation or the moving image contour coding using transformation and motion estimation guarantees excellent reconstruction even with a small amount of generated bit.

Abstract: A sequential polygon approximation apparatus for a contour and a method thereof which are capable of sequentially extracting a vertex in the order of a contour pixel when performing a polygon approximation so as to code a shape information, which includes an error computation unit for obtaining a contour pixel between two contour pixels of the vertex extraction unit from the contour pixel memory and computing an error which is a distance between a straight line between two contour pixels and a pixel from the contour pixel memory, a comparison unit for comparing an error obtained by the error computation unit with a threshold value, and a controller for controlling the vertex extraction unit and the error computation unit in accordance with the comparison result of the comparison unit, setting the pixel just before the final pixel of a first region as a first vertex when the error obtained in the first region which is the region of the contour pixel sequentially determined for the vertex extraction from a fixe

Abstract: The stroke code for encoding image partitions is disclosed. An image partition may be derived from a natural image via an image segmentation process or may be directly available from map colored images such as commercially available clip art. The stroke code is a language for developing an image partition using three-way chains. Each stroke consists of a starting location, and one or two boundary chains. Each chain of a stroke is terminated by encounters with previously drawn separators or the image boundary. Not all encounters with previously drawn separators terminate a stroke chain. Further information disambiguates each encounter. The stroke language largely retains a three way decision character for each stroke chain and codes each boundary separator only once. Preferred embodiments of a stroke language encoder and a stroke language decoder are disclosed. Stroke location and termination decisions are conditioned with simple context models derived from previously drawn strokes.

Abstract: A precompression extrapolation method extrapolates image features of arbitrary configuration to a predefined configuration to facilitate compression of the image features. This extrapolation method allows compression of arbitrary image features to be performed in a conventional manner, such as discrete cosine transform or lattice wavelet compression. This method includes defining an extrapolation block boundary about an object or image feature that is bounded by a feature boundary. Pixels within the block, but not included in the image feature, are identified as non-feature pixels. The feature pixels in the feature boundary include pixel values that correspond to the image characteristics of the pixels. The non-feature pixels adjacent the pixels of the feature boundary are assigned pixel values that include the pixel values of the adjacent feature boundary pixels.

Abstract: Bit streams transmitted from a coding system through a transmission channel are divided on a block-by-block basis based on bit numbers thereof and a parity bit added to each bit stream is checked to decide whether or not each bit stream is error bit stream having one or more error bits therein. Based on the check result, either each bit stream or a proximate bit stream for each bit stream is selected as an optimum bit stream, wherein the proximate bit stream is generated based on degrees of proximity between masked transform coefficients for each candidate bit stream and masked transform coefficients for each of reference bit streams spatially adjacent to each bit stream.

Abstract: An information inputting method, as well as an information inputting device, extracts effective characteristic points from inputted stroke information and outputs coordinate information relating to the extracted characteristic points. First point coordinates values are input in a sampling cycle at a predetermined sampling rate. Second point coordinates values are input in the sampling cycle delayed a predetermined time behind the inputting of the first point coordinate values. A first vector is formed based on the first and second point coordinate values. First and second point coordinate values are also input in the subsequent sampling cycle, and a second vector is determined based thereon. The first point coordinate values inputted earlier are selected when these values cause the angle formed between the first and second vectors to exceed a predetermined threshold value. Information concerning the selected first point coordinates values are output to, for example, a host device.

Abstract: A contour encoding method codes a current contour in a current frame based on a predicted current contour which is selected among previously reconstructed contours within a previous frame. Once the predicted current contour which is most similar to the current contour is detected, the method matches the predicted current contour with the current contour and widens the predicted current contour to thereby detect non-matching sections and matching sections between the current contour and the predicted current contour. Subsequently, the portions of the current contour which correspond to the non-matching sections are encoded through a polygonal approximation technique. The result of the approximation technique and information for the predicted current contour are multiplexed and then transmitted as encoded contour data.

Abstract: A character processing apparatus for recognizing coordinate information expressing a character, an exterior outline, an interior outline, a horizontal line and a vertical line; classifying the coordinate information expressing a character, the exterior outline, the interior outline, the horizontal line and the vertical line, and then determining a pair of the horizontal line and the vertical line; and setting a skipping order at the time of low pixel so as to be converted into character data for a bit map development composed of a control point coordinate for expressing the outline of the subject character, band information and coordinate value information about the control point which is not included in the band information.

Abstract: An image encoding method for encoding moving images, comprising the steps of:detecting characteristic points of a moving image;encoding a chain generated based on information relating to the characteristic points in such a manner as to generate chain encoded data, with the chain corresponding to an outline of the image;detecting motion of the chain and generating information corresponding to the motion; andmultiplexing the chain encoded data and the motion information.

Abstract: A method for encoding a video signal of a contour of an object based on the iterated refinement technique determines a plurality of vertices on the contour; and calculates a contour pixel bit-number required to encode all the contour pixels and a vertex bit-number needed to encode all the vertices. After generating a determination signal representing the smaller of the contour pixel bit-number and the vertex bit-number, the method encodes contour information representing the vertices or the contour pixels based on the determination signal to generate a coded data.

Abstract: A segment-based contour encoding method is provided. A motion compensation of an object in a current image with respect to a reference image is performed to obtain a motion-compensated image. The contour contained in the image motion-compensated is compared with a contour of an object in a current image, to divide the whole contour of the object in the current image into a global motion compensated (GMC) segments and a global motion compensation failure (GMF) segments. Each GMF segments is divided into a local motion compensated (LMC) segment and a local motion compensation failure (LMF) segment. The GMC segments and the LMC segments are motion-compensation-encoding and the LMF segments are encoded using only pixel information constituting each LMF segment. The contour encoding method can reduce the number of the LMF segments and can perform an efficient encoding operation for motion compensation encoding.

Abstract: An apparatus for encoding a contour of an object in a video signal by using a baseline-based shape coding technique first determines a baseline based on contour image data and generates a one-dimensional sample list of the contour based on the baseline. Subsequently, the apparatus produces a reconstructed contour based on the one-dimensional sample list and extracts the difference between the reconstructed contour and the original contour. The detected difference is sampled in order to create an error sample list and, finally, the one-dimensional sample list is compensated based on the error sample list.

Abstract: A system and method for efficiently specifying vertex information for a three-dimensional graphical object which includes a plurality of geometric primitives. The method comprises organizing a first subset of the object's vertices into a strip of geometric primitives (typically triangles). The method next includes representing vertices of the strip by encoding a plurality of commands into a data stream. These commands are usable, during decompression, to reassemble the strip of primitives from a list of vertices. Selected commands specify that attributes of a particular vertex (position, color, normal value, etc.) are to be stored into a mesh buffer for use in forming subsequent primitives. The mesh buffer includes a fixed maximum number of memory locations which are accessible during the decompression process. In one embodiment, vertex parameters are "pushed" on to the mesh buffer, which is organized as a stack.

Abstract: In an image signal coding and decoding method in which feature points in an image signal are detected and information concerned with sets of continuous feature points is chain-coded, the quantity of information is reduced without deterioration of image quality. An image signal coding and decoding method in which components of a direction displacement between the direction of transition from a feature point two feature points before a current feature point to a preceding feature point and the direction of transition from the preceding feature point to the current feature point are detected and subjected to entropy coding.

Abstract: The invention relates to an image sequence coding method in which images are segmented and coded with respect to their contours and textures. The texture coding step is carried out by means of a new technique relying on a wavelet decomposition of the images, called quincunx bidimensional wavelet transform and adapted to a region-based coding scheme, for applications allowing to reach very low bit rates while keeping a good image quality.

Abstract: A method for encoding a contour of an object decides a plurality of primary vertices on the contour based on a polygonal approximation technique and also determines a multiplicity of secondary vertices on the contour which are disposed at one of a predetermined set of locations away from a previously determined secondary vertex. Then, the method encodes position information of the primary vertices to provide encoded primary vertex data and codes position information of the secondary vertices to provide coded secondary vertex data. Finally, either the encoded primary vertex data or the coded secondary vertex data is selected as encoded contour data in response to a selection signal which is generated by estimating an encoding error with a threshold TH and the encoded contour data is provided to a transmitter for the transmission thereof.

Abstract: A method for encoding a video signal of a contour of an object based on an iterated refinement technique determines a plurality of primary vertices on the contour; and calculates a displacement for each contour segment to produce displacements on all pairs of two adjacent primary vertices. After determining a dynamic range as a criterion for encoding pixel data on the contour based on the displacements, the method searches a larger contour segment which is a contour segment whose number of bits required to encode pixel data thereon is larger than that of the dynamic range, and inserts one or more secondary vertices on the larger contour segment based on the dynamic range.

Abstract: A picture encoding apparatus and the method in which the quantization step width of the lower hierarchy data having a resolution higher than that of the upper hierarchy data is determined for each predetermined block of respective hierarchy data based on the quantization step width determined by the upper hierarchy data having a low resolution, so that the additional code indicating the characteristics of quantizer can be omitted thereby the compression efficiency can be improved and the deterioration of picture quality can be reduced when picture data is hierarchical-encoded.

Abstract: Previous vertices on a previous contour are mapped onto a current contour in this method to form a predicted vertex on the current contour corresponding to each of the previous vertices and vertex displacements between the previous vertices and their corresponding predicted vertices are calculated. After current vertices on the current contour being determined, the current contour is coded based on either the current vertices to thereby provide intra-coded data or the previous contour to provide inter-coded data; and either the intra-coded data or the inter-coded data is selected based on the vertex displacements as encoded data for the current contour.

Abstract: In an encoding method for use in an encoder to encode an image signal including an object having object pixels, a background having background pixels and a contour signal representing the contour of the object, the contour signal is first encoded based on mask data having first and second binary values to designate an object pixel and a background pixels, respectively and then decoded to generate a decoded contour. In the mean time, a primary padded image signal obtained by applying a primary padding on the image signal is divided into multiplicity of equal-sized primary padded blocks and then each of the primary padded blocks is defined either as a contour block or as an object block based on the decoded contour to thereby select contour blocks and object blocks.

Abstract: A contour of an object in an image having one or more objects therein is reconstructed based on encoded contour data for the contour and a previously reconstructed contour for an object included in the image. The method, first, decodes the encoded contour data for a current contour to thereby provide a reconstructed current contour and checks whether the reconstructed current contour overlaps with any previously reconstructed contours. If none of the previously reconstructed contours overlaps with the reconstructed current contour, the method sets the reconstructed current contour as a currently reconstructed contour. If, otherwise, the reconstructed current contour overlaps with a previously reconstructed contour, the reconstructed current contour is modified not to overlap with the previously reconstructed contour and the method sets the modified contour as the currently reconstructed contour.

Abstract: A contour image coding method divides a video frame into a multiplicity of blocks and detects contour blocks and determines a plurality of vertices on each contour segment included in each contour block and codes the contour image based on the vertices on each contour segment for each contour block.

Abstract: An apparatus, capable of improving the overall coding efficiency by encoding an image signal having a still object based on an improved object oriented coding technique, comprises a first detector for detecting a contour image signal placed at the boundary of the object within the image signal and for encoding the detected contour image signal, to thereby generate a first encoded image signal; a divider for dividing the image signal into a plurality of image blocks with a predetermined identical size; a second detector for detecting image blocks containing the contour image signal from the image blocks and for generating a reconstructed image block for each of the detected image blocks; an encoder, responsive to a selection signal corresponding to the type of the image blocks, for selectively encoding the reconstructed image blocks or the non-detected image blocks, to thereby produce a second encoded image signal; and a formatter for formatting the first and the second encoded image signals.

Abstract: The invention relates to an image sequence coding method in which images are segmented and coded with respect to their contours and textures. The texture coding step is carried out by means of a lattice vector quantization method adapted to region-based coding schemes where regions have arbitrary shapes, for applications allowing to reach very low bit rates while keeping a good image quality.

Abstract: A method encodes a current contour of an object in a video signal based on a previous contour thereof by extending the current contour over neighboring pixels thereto. In the encoded method, a displacement between the extended and the previous contours is found to set the displacement as a motion vector for the current contour. The previous contour is overlapped onto the extended contours based on the motion vector to thereby provide an overlapped previous contour and an overlapped extended contour. A mismatch between the overlapped contours is detected and encoded to thereby generate encoded error data, wherein, the encoded contour data includes the encoded error data and the motion vector.

Abstract: A method and apparatus for carrying out rapid, block-based SAD search operations for facial ellipses in a CIF videophone image frame are disclosed. A set of search templates, each having an ellipse pattern with two concentric, contiguous perimeters is defined and a set of parameter values are calculated off line for each template: a predetermined subfactor, index number, number of pixels in its dithered perimeter pattern, and first and second match thresholds. The pixels in the lower portion of each perimeter in the pattern are thinned by dithering so as to emphasize the upper contour of the ellipse. Values of +2k for interior perimeter pixels and -2k for exterior pixels in the template are then added to corresponding +k edge pixels and -k non-edge pixels in each search position of the template on a thresholded binary edge map of the image frame.