CHAPTER 25.   README.TXT
25.1   Software Requirements
25.2   Install the Software
25.3   Image Recognition
25.3.1   Image Processing
25.3.2   BioFilter Templates
25.3.3   BioFilter Training and Matching
25.3.4   Neural Filters
25.3.5   NeuralNet Filter
25.3.6   Batch File



Chapter 25.   Readme.txt

25.1   Software Requirements

Software Requirements:

(1) Windows .Net Framework 1.1.
(2) J# .Net Redistributable.
(3) Internet Explorer.

(1) To get the latest version .Net Framework 1.1, use the Internet Explorer, then click
Tools\Windows Update".

(2) To get J# .Net Redistributable, either get it from this Microsoft site directly:
http://www.microsoft.com/downloads/details.aspx?familyid=E3CF70A9-84CA-4FEA-9E7D-7D674D2C7CA1&displaylang=en
or get it from the CD with the following path: CD:\vjredist.exe. Please install it by double clicking it.

25.2   Install the software

1. If you have not done so, go to Internet Explorer, then click  �Tools\Windows Update" to download Windows .Net Framework 1.1.

2. Click CD:\vjredist.exe to install Microsoft J# .Net Redistributable.

3.Click CD:\setup.exe to install the ImageFinder.

25.3   Image Recognition

The ImageFinder is an Image Recognition software. At a minimum, you will need the following steps:

These steps can be combined into a single batch click.

Image Processing sets up 3 image filters: Edge Filters, Threshold Filters, and Clean Up Filters.

Feature Space Recognition trains two filters, the BioFilter and NeuralFilter, and uses these filters for 1:N and N:N Image Matching.

Input Space Recognition trains the Neural Net Filter and makes 1:N or N:N Matching at the pixel level. The output of the Feature Space Recognition is usually the input of the Neural Net Filter.

25.3.1   Image Processing

If you are not familiar with Image Processing, please try the following:

Setting 1:

Edge Filters: Sobel 1 (or Sobel 2)
Threshold Filters: Dark Background 128
Setting 2:
Edge Filters: None (�Enter An Edge Filter�)
Threshold Filters: Light Background 128
Other settings can be adopted by trial and error. You should choose an image pre-processing procedure where the sample objects stand out, otherwise change the options.

25.3.2   BioFilter Templates

BioFilter matching will have a few steps:

Initialization sets the ImageFinder parameters. To start with, simply use the default setting.

To converting Images to Records in a feature space, there are two steps:

Similar steps can be performed for file input.

25.3.3   BioFilter Training and Matching

The main task of training is to prepare a file match.txt, which is a list of matching pairs. Click �BioFilter\Train (match.txt required)� to train the BioFilter.

For BioFilter N:N Matching: click �BioFilter/BioFilter N:N Match (Trained)�. For BioFilter 1:N Matching: click �BioFilter/BioFilter 1:N Match (Trained)�.

The result file contains many blocks: each image has a block.  Line 1 in each block is the input and the rest of the lines are output.

25.3.4   Neural Filters

The Neural Filter runs parallel to the BioFilter.

25.3.5   Neural Net Filter

Neural Network matching will have a few steps:

Initialization sets the ImageFinder parameters. To start with, simply use the default setting. Training and matching take the following steps:
1. Enter key-segments into the ImageFinder (keys are used to teach this software what to look for);
2. Click the NeuralNet/Train to teach the software what to look for.
3. Save all the images you want to look through into a directory (search-directory) and enter it into the software;
4. Click a NeuralNet/Search --- the computer is now looking through the images.
5. The output is a web page, which is a list of names and weights (scores):
25.3.6   Batch File

Batch commands allow you to save your setting and to execute your problem in a few clicks:

Click Batch/Save to save your setting; later click Batch/Load to load your setting and click Batch/Run to run your problem.
 
 

Return