文件列表:

ColorImageSegmentation\felzenszwalb.exe (110592, 2007-02-23)
ColorImageSegmentation\felzenszwalb (190727, 2006-05-07)
ColorImageSegmentation\segmentation_ui.m (13773, 2007-04-30)
ColorImageSegmentation\getHierarchicalLabel.m (970, 2007-04-23)
ColorImageSegmentation\countleadingblanks.m (448, 2007-04-23)
ColorImageSegmentation\deleteduplicates.m (626, 2007-03-27)
ColorImageSegmentation\segmentation_ui.asv (13494, 2007-03-27)
ColorImageSegmentation\countsegments.m (493, 2007-03-22)
ColorImageSegmentation\loaddir.m (580, 2007-03-22)
ColorImageSegmentation\loadstrings.m (622, 2007-03-22)
ColorImageSegmentation\segmborders.m (616, 2007-03-22)
ColorImageSegmentation\visualize_segm2.m (693, 2007-06-27)
ColorImageSegmentation\randomcolormap.m (709, 2007-03-22)
ColorImageSegmentation\segmentation_ui.fig (4191, 2007-06-28)
ColorImageSegmentation\categories.txt (2043, 2007-03-02)
ColorImageSegmentation\convert2_8bit.m (399, 2007-02-27)
ColorImageSegmentation\Starting.m (167, 2007-06-11)
ColorImageSegmentation\Starting.asv (80, 2007-06-11)
ColorImageSegmentation\temp.ppm (294927, 2007-06-28)
ColorImageSegmentation\015.jpg (38325, 2007-04-14)
ColorImageSegmentation\test.m (5021, 2007-06-26)
ColorImageSegmentation\test.fig (2580, 2007-06-26)
ColorImageSegmentation\test.asv (4766, 2007-06-26)
ColorImageSegmentation (0, 2007-08-19)

semi-automatic image segmentation & annotation ============================================== Copyright 2007 AUTHOR: Steffen Gauglitz University of California at Santa Barbara, Vision Research Lab contact: Prof. Manjunath, manj@ece.ucsb.edu Modification/redistribution granted only for the purposes of teaching, non-commercial research or study. If you use this application we would appreciate it if you cite: Elisa Drelie Gelasca, Joriz De Guzman, Steffen Gauglitz, Pratim Ghosh, JieJun Xu, Amir M. Rahimi, Emily Moxley, Zhiqiang Bi, B. S. Manjunath: "CORTINA: Searching a 10 Million+ Images Database", submitted to the 33rd International Conference on Very Large Data Bases (VLDB), September 23-28 2007, University of Vienna, Austria. OVERVIEW REQUIREMENTS DESCRIPTION INSTALLATION RUNNING THE GUI INPUT SPECIFICATIONS OUTPUT SPECIFICATIONS USER INTERFACE == OVERVIEW == This is a set of MATLAB scripts for semi-automatic image segmentation & annotation. It uses a segmentation algorithm to preprocess the image and allows the user to recombine, refine and create segments and to label them out of a set of given categories. == REQUIREMENTS == This tool was successfully tested under Linux and MS Windows. It requires ImageMagick to convert image formats, please install it from http://www.imagemagick.org and make sure your system can call the 'convert' command. == DESCRIPTION == As a preprocessing step, this tool uses the segmentation algorithm described in: Efficient Graph-Based Image Segmentation Pedro F. Felzenszwalb and Daniel P. Huttenlocher International Journal of Computer Vision, 59(2) September 2004. A compiled version for Linux and MS Windows is included, however, the source code can be obtained online: http://people.cs.uchicago.edu/~pff/segment/ == INSTALLATION == First make sure that your OS knows how to find the segmentation executable (felzenszwalb or felzenszwalb.exe, respectively) by putting it somewhere in the OS's search path. Now make sure MATLAB can find all of the scripts that are included by either making their directory the current directory or including it in the MATLAB path. == RUNNING THE GUI == Run the GUI by executing the following command in MATLAB: segmentation_ui( , ); See 'help segmentation_ui' for explanation of the two parameters. The other scripts are helper functions for the GUI. Feel free to explore by typing 'help '. == USER INTERFACE == After starting the GUI (see above), it presents the first image in a pre-segmented version. Browse the image list with the "< prev"/"<< -10"/"+10 >>"/"next >" buttons. You can (de)select segments by just clicking them (selection accumulates) or using the "sel none"/"sel all"/"sel invert" buttons. Clicking "refine sel" refine the selected segments (up to four times) by applying the segmentation algorithm with lower thresholds, "merge sel" merges all selected segments into one segment. "polygon" allows you to create a new segment as an arbitrarily shaped polygon. "median filter" smoothens the segmentation (you will be asked to enter the kernel size). Finally, "label sel" lets you assign a label to the selected segment(s). You can also select a previously labeled segment by clicking its label in the list on the right side. Click on "save" to save this image's results: - the segmentation mask will be saved as .seg.gif - the labels will be saved as .label.mat == INPUT SPECIFICATIONS == You may use any image format MATLAB and ImageMagick can handle. Define your categories by listing them in a raw text file. You can create a hierarchy by intending with spaces, so for example for a category file like this: animal tiger lion objects man made object desk chair computer a segment showing a 'desk' gets labeled as 'object\man made object\desk'. == OUTPUT SPECIFICATIONS == Segmentation results are stored as files named .seg.gif. Each color in the GIF image denotes a segment. (The actual color is meaningless, so we better refer to them as 'tokens'). Segmentation labels are stored in .label.mat, containing a matlab struct. If you load that struct and diplay it, you will see something like this: labels = 1x2 struct array with fields: segmenttoken label labels(1) = segmenttoken: 0 label: 'animal\tiger' ...where the '0' refers to the token in the GIF image denoting the area of the 'tiger'. Files with extension .ppm are preprocessing results. They may safely be deleted after work has been concluded.

近期下载者：

相关文件：

评论：[我要评论] [举报此文件]

收藏者：