ShapeLogic menu

• CapitalLetterMatch
• Segment
• ParticleCounter
• Color Particle Analyzer
• RGB Color Particle Analyzer
• Stream Vectorizer
• Color Replacer

CapitalLetterMatch

Main command for letter recognition. This is a macro that:

• Turns an image into a black and white image
• Skeletonizes an image, turning line into thin lines
• Runs the Stream Vectorizer plugin described below

Usage instruction

Segment

Segment is a 4 connect color segmentation algorithm. Currently the color distance can only be set in the code, but an input dialog may be added in the future.

ParticleCounter

This will count particles in a color image but cannot do any edge tracing or analysis. The particle analyzer (see below) can also be called with a count only option and has the following advantages over the particle counter:

• Set the background or foreground color
• Print out internal information
• Turn image into a black and white mask

Usage instruction

Color Particle Analyzer

Color Particle Analyzer is a plugin to analyze color or black and white images. The background color is found automatically.

Usage instruction

RGB Color Particle Analyzer

RGB Color Particle Analyzer is a plugin to analyze color or black and white images. The background or foreground color is set by the user.

Usage instruction and co-localization images.

Stream Vectorizer

Main Java code for letter recognition. It is a plugin that takes a line image with thin or skeletonized lines and vectorizes and analyzes them then matches them to letters.

Based on:

• Lazy stream declarative programming
• Language to describe annotations and locations
• Max Distance Vectorizer

Usage instruction

Color Replacer

Little utility plugin to isolate color or gray interval. You select a color and change all the pixels in the image that are close to this into that color. You can also turn it into a black and white mask.

Usage instruction

DigiMatch

Main command for digit recognition. It is a macro similar to the CapitalLetterMatch but for digits, it calls the digit match plugin Digi Match Stream Vectorizer.

Usage instruction

Line Vectorizer

First approach to vectorizing lines. Based on splitting lines into pieces that are 5 pixels long. This was pretty unstable.

Direction Based Vectorizer

Second approach to vectorizing lines. Based on special handling of the intersection points. More stable.

Max Distance Vectorizer

Third approach to vectorizing lines. Standard approach to vectorizer trace full line between intersection and split it up based on max distance from a line between the end points. This is stable. This implementation only handles the first found polygon. Same approach that is used in Stream Vectorizer.

Find color hypothesis

This is the algorithm that is used to find the background in the Color Particle Analyzer.

StraightCapitalLetterMatch

First very crude OCR, that can only recognize one straight capital letter. Based on

• Goal driven declarative programming
• Direction Based Vectorizer

Vectorize Information

Does the same as Stream Vectorizer, but also writes out all the internal information that is used in the process:

• Points
• Multi lines
• Polygons
• Annotations for all of these

Usage instruction

Digi Match Stream Vectorizer

Main Java code for digit recognition. This is a plugin to take a line image with thin or skeletonized lines and vectorize and analyze them, then match them to digits.

Usage instruction

TaskBasedCapitalLetterMatch

Second crude OCR, that can recognize one capital letter. Based on

• Goal driven declarative programming
• Language to describe annotations and locations
• Max Distance Vectorizer