1 MSEGMENT Segments an image with thresholding and the fuzzy c-Means technique. SYNOPSIS $ msegment [ options ... ] input_file output_file 2 DESCRIPTION segment segments an image by analyzing the histograms of the color components and identifying units that are homogeneous with the fuzzy c-means technique. The scale-space filter analyzes the histograms of the three color components of the image and identifies a set of classes. The extents of each class is used to coarsely segment the image with thresholding. The color associated with each class is determined by the mean color of all pixels within the extents of a particular class. Finally, any unclassified pixels are assigned to the closest class with the fuzzy c-means technique. The fuzzy c-Means algorithm can be summarized as follows: o Build a histogram, one for each color component of the image. o For each histogram, successively apply the scale-space filter and build an interval tree of zero crossings in the second derivative at each scale. Analyze this scale-space ``fingerprint'' to determine which peaks or valleys in the histogram are most predominant. o The fingerprint defines intervals on the axis of the histogram. Each interval contains either a minima or a maxima in the original signal. If each color component lies within the maxima interval, that pixel is considered ``classified'' and is assigned an unique class number. o Any pixel that fails to be classified in the above thresholding pass is classified using the fuzzy c-Means technique. It is assigned to one of the classes discovered in the histogram analysis phase. The fuzzy c-Means technique attempts to cluster a pixel by finding the local minima of the generalized within group sum of squared error objective function. A pixel is assigned to the closest class of which the fuzzy membership has a maximum value. For additional information see Young Won Lim, Sang Uk Lee, "On The Color Image Segmentation Algorithm Based on the Thresholding and the Fuzzy c-Means Techniques", Pattern Recognition, Volume 23, Number 9, pages 935-952, 1990. 2 EXAMPLES To segment an image of a cockatoo, use: $ msegment cockatoo.miff cockatoo.segmented To segment a raw grayscale image, use: $ msegment -size 768x512 gray:raw image.miff 2 OPTIONS -cluster_threshold value eliminate clusters that are insignificant. The number of pixels in each cluster must exceed the the cluster threshold to be considered valid. The default is 1%. -colorspace value the type of colorspace: GRAY, OHTA, RGB, XYZ, YCbCr, YIQ, YPbPr, or YUV. Color reduction, by default, takes place in the RGB color space. Empirical evidence suggests that distances in color spaces such as YUV or YIQ correspond to perceptual color differences more closely than do distances in RGB space. These color spaces may give better results when color reducing an image. Refer to quantize(9) for more details. -comment string annotate an image with a comment. By default, each image is commented with its file name. Use this option to assign a specific comment to the image. Optionally you can include the image filename, type, width, height, or scene number by embedding special format characters. Embed %f for filename, %m for magick, %w for width, %h for height, %s for scene number, or \n for newline. For example, -comment "%m:%f %wx%h" produces an image comment of MIFF:bird.miff 512x480 for an image titled bird.miff and whose width is 512 and height is 480. If the first character of string is @, the image comment is read from a file titled by the remaining characters in the string. -compress type the type of image compression: QEncoded or RunlengthEncoded. Specify +compress to store the binary image in an uncompressed format. The default is the compression type of the specified image file. -density x vertical and horizontal density of the image. This option specifies an image density for a Postscript page. The default is 72 dots per inch in the horizontal and vertical direction. Use this option to alter the default density. -display host:display[.screen] specifies the X server to contact; see X(1). -font name This option specifies the font to be used for displaying normal text. The default is fixed. -interlace type the type of interlacing scheme: NONE, LINE, or PLANE. This option is used to specify the type of interlacing scheme for raw image formats such as RGB or YUV. NONE means do not interlace (RGBRGBRGBRGBRGBRGB...), LINE uses scanline interlacing (RRR...GGG...BBB...RRR...GGG...BBB...), and PLANE uses plane interlacing (RRRRRR...GGGGGG...BBBBBB...). -label name assign a label to an image. Use this option to assign a specific label to the image. Optionally you can include the image filename, type, width, height, or scene number in the label by embedding special format characters. Embed %f for filename, %m for magick, %w for width, %h for height, or %s for scene number. For example, -label "%m:%f %wx%h" produces an image label of MIFF:bird.miff 512x480 for an image titled bird.miff and whose width is 512 and height is 480. If the first character of string is @, the image label is read from a file titled by the remaining characters in the string. When converting to Postscript, use this option to specify a header string to print above the image. -matte store matte channel if the image has one. -page x{+-}{+-} preferred size and location of the Postscript page. Use this option to specify the dimensions of the Postscript page in picas or a TEXT page in pixels. The default for a Postscript page is to center the image on a letter page 612 by 792 dots per inch. The left and right margins are 18 picas and the top and bottom 94 picas (i.e. 612x792+36+36). Other common sizes are: 540x720 Note 612x1008 Legal 842x1190 A3 595x842 A4 421x595 A5 297x421 A6 709x1002 B4 612x936 U.S. Foolscap 612x936 European Foolscap 396x612 Half Letter 792x1224 11x17 1224x792 Ledger The page geometry is relative to the vertical and horizontal density of the Postscript page. See -density for details. The default page dimensions for a TEXT image is 612x792+36+36. -quality value JPEG quality setting. Quality is 0 (worst) to 100 (best). The default is 85. -scene value image scene number. -size {%}x{%}{+colors}{!} width and height of the image. Use this option to specify the width and height of raw images whose dimensions are unknown such as GRAY, RGB, or CMYK. In addition to width and height, use -size to tell the number of colors in a MAP image file, (e.g. -size 640x512+256). -smoothing_threshold value eliminate noise in the second derivative of the histogram. As the value is increased, you can expect a smoother second derivative. The default is 5.0. -verbose print detailed information about the image. This information is printed: image scene number; image name; converted image name; image size; the image class (DirectClass or PseudoClass); the total number of unique colors; and the number of seconds to read and convert the image. Options are processed in command line order. Any option you specify on the command line remains in effect until it is explicitly changed by specifying the option again with a different effect. Change '-' to '+' in any option above to reverse its effect. For example, specify +matte to store the image without its matte channel. By default, the image format is determined by its magic number. To specify a particular image format, precede the filename with an image format name and a colon (i.e. ps:image) or specify the image type as the filename suffix (i.e. image.ps). See convert(1) for a list of valid image formats. When you specify X as your image type, the filename has special meaning. It specifies an X window by id, name, or root. If no filename is specified, the window is selected by clicking the mouse in the desired window. Specify input_file as - for standard input, output_file as - for standard output. If input_file has the extension .Z or .gz, the file is uncompressed with uncompress or gunzip respectively. If output_file has the extension .Z or .gz, the file size is compressed using with compress or gzip respectively. Finally, precede the image file name with | to pipe to or from a system command. Use an optional index enclosed in brackets after a file name to specify a desired subimage of a multi-resolution image format like Photo CD (e.g. img0001.pcd[4]). If output_file already exists, you will be prompted as to whether it should be overwritten. 2 ENVIRONMENT display To get the default host, display number, and screen. 2 COPYRIGHT Copyright 1994 E. I. du Pont de Nemours and Company Permission to use, copy, modify, distribute, and sell this software and its documentation for any purpose is hereby granted without fee, provided that the above copyright notice appear in all copies and that both that copyright notice and this permission notice appear in supporting documentation, and that the name of E. I. du Pont de Nemours and Company not be used in advertising or publicity pertaining to distribution of the software without specific, written prior permission. E. I. du Pont de Nemours and Company makes no representations about the suitability of this software for any purpose. It is provided "as is" without express or implied warranty. E. I. du Pont de Nemours and Company disclaims all warranties with regard to this software, including all implied warranties of merchantability and fitness, in no event shall E. I. du Pont de Nemours and Company be liable for any special, indirect or consequential damages or any damages whatsoever resulting from loss of use, data or profits, whether in an action of contract, negligence or other tortuous action, arising out of or in connection with the use or performance of this software. 2 ACKNOWLEDGEMENTS Andy Gallo, University of Delaware. This software is strongly based on software Andy wrote. 2 AUTHORS John Cristy, E.I. du Pont De Nemours and Company Incorporated