人工神经网络模式识别系统

The programs in this directory illustrate geometric feature detection as discussed in chapter6. The features detected are intersections, end points and loops. As a preliminary step in the feature detection process, thinning(see chapter 3) is performed. The two programs found in this directory are: 1) showdata 2) feature Sample data can be found in: chrdata.bm Sample output data can be found in: chrdata.0, chrdata.1,...,chrdata.9. SHOWDATA: The showdata program read character data in the format of a ".BM" file, converts it to a matrix representation and outputs the image in text mode to the screen. Characters are on a 16x 24 pixel grid. ".BM" consist of an arbitrary number of records where each record consists of: byte 0x00: always 0x00 byte 0x01: ascii code of the character bytes 0x02-0x31: data in 1 bit-per-pixel format. (i.e. Thus it takes 2 bytes to represent a row of the character matrix.) FEATURES: Accepts a ".BM" file as input and creates: 1) a thinned version of the target character and 2) a version of the character consisting only of pixels that are loop members. The original, thinned and loop member versions are all displayed. Coordinates for intersections and endpoints are calculated and displayed. Loop mass and the coordinates of the loop center of mass are calculated and displayed as [mass,(x,y)]. TO Compile: icc showdata.cpp <enter> Usage: showdata SourceFile <enter> where SourceFile is the name of a ".BM" format file (both name and extention are required) FEATURES: To compile: icc features.cpp <enter> Usage features SourceFileName <enter> x where SourceFilename is the name of a ".BM" format file. (ONLY the file name should appear on the command line as the program will fill in the .BM extention. Output from the program will go to both the screen and to files SourceFilename.1, SourceFilename.2, SourceFilename,...