1.   INTRODUCTION
1.1   What will the ImageFinder do?
1.2   Biometrics
1.3   Software Requirements
1.4   Customized Software
1.5   Trails

1.   Introduction

 The Attrasoft ImageFinder looks at a sample image or several images and will match all similar images from a directory or a file.

Figure 1.1   ImageFinder.

Attrasoft provides Universal Image Identification and Retrieval Technology, which is

Image Recognition technology will increasingly replace human labor in the form of: which will translate into saving man-hours, saving money, making more money with fewer resources, doing new things, and/or shorten service turnaround time.

When the Attrasoft ImageFinder learns an image(s) and goes to a directory/file to retrieve similar images, it does not deal with keywords. This software learns the content of an image or several images directly from the image(s) and retrieves all similar images based on the content. The Attrasoft ImageFinder provides the users with a tool for image matching.

The central task in any image data management system is to retrieve images that meet some specified constraints. The Attrasoft ImageFinder provides users with a tool for content-based image retrieval.

 Applications are limited only by your imagination:

1.1   What will the ImageFinder do?

Attrasoft ImageFinder can match images (jpg or gif) which:

 Key-images, or Key-segments are used to tell the ImageFinder what to look for. This is called training. After training, the ImageFinder is ready to retrieve all similar images. This includes all:


1.2   Biometrics

Biometrics use human faces, fingerprints, voice, iris, �,  to verify or to identify a person.

There are three generally accepted methods for performing human Verification or Identification. These are based on:

a. Something the user KNOWS (such as a password);
b. Something the user POSSESSES (such as a card/badge, called tokens); or
c. Something the user IS (a physical characteristic, or biometric, such as a fingerprint)
However, passwords can be compromised in many ways - they can be forgotten, written down, guessed, stolen, "cracked", or shared. Tokens, like a telephone card or credit card, can be lost, forgotten, stolen, given away, or duplicated.

Biometrics Verification or Identification can be accomplished by measurement of a unique biological or behavioral feature of the user to verify identity through automated means.

To determine if one image sample "matches" another image sample, they must be compared using a unique algorithm. Generally, the result of this comparison is a "score", indicating the degree to which a match exists. This score is then compared to a pre-set threshold to determine whether or not to declare a match. The comparison is performed using the captured image and previously stored images.

There are several types of image matching:

Verification

Verification is a one-to-one (1:1) Matching of a single sample set (biometric identifier record) against another. Generally, the first sample is newly captured and the second is the enrolled identifier on file for a particular subject. In a user authentication environment, a score exceeding the threshold would return a 'match', resulting in the authentication of the user. A score below the threshold would return a 'no-match', resulting in the denial of access.


Identification

Identification is a one-to-many (1:N) Matching of a single sample set against a database of samples, with no declared identity required. The single image is generally the newly captured sample and the database contains all previously enrolled samples. Scores are generated for each comparison, and an algorithm is used to determine the matching record, if any. Generally, the highest score exceeding the threshold results in Identification.


Search or Retrieval

Search is similar to Identification, i.e. 1:N Matching; however, the result is a set of possible matching images, not a classification. "Identification" returns a classification, while "Search" returns multiple matched images.


Classification

In the above three cases; only one class of image(s) is compared with a set of existing images.  "Classification" is N:1 or N:N Matching.


1.3   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 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 this CD with the following path: CD:\vjredist.exe. Please install it by double clicking it.

1.4   Customized Software

There are hundreds of parameters in the software. The fine-tuning of these hundreds of parameters via Customization is Attrasoft's expertise. This software has several examples of Customized Software:

(1) FaceFinder;
(2) DocumentFinder;
(3) ImageExaminer.

As you will see, Customized Software has no parameters to adjust.

If you need a customized version of the ImageFinder, please contact imagefinder@attrasoft.com.

Customized versions can accommodate:
 


1.5   Trails

This User�s Guide provides specific guidance for the software. This Guide is organized as follows:

Chapter 2 gives an Overview of the Software and Overview of the Chapters.

Part I introduces Customized Software. Part I includes:

Part II introduces the Basic Architecture:
Image Preprocessing Normalization Feature Space Recognition Pixel Level (Input Space) Recognition
Part II includes:
 Chapter 6,  Image Processing;
 Chapter 7,  BioFilter;
 Chapter 8,  NeuralFilter;
 Chapter 9,  NeuralNet Filter or ABM Filter,
 Chapter 10, Parameters; and
 Chapter 11, Batch.
Part III introduces more topics on ABM Filters. Part II includes:
 Chapter 12, ABM Filter Short-Search;
 Chapter 13, ABM Filter Long-Search;
 Chapter 14, ABM Filter Parameter Advisor;
 Chapter 15, Biometrics;
 Chapter 16, Segment-Locator; and
 Chapter 17, Customized Point-Locator.
Part IV introduces Extended Architecture. Part IV includes:
 Chapter 18, BioFilter II;
 Chapter 19, Neural Filter II;
 Chapter 20, ABM Filter II.
Part V discusses topics interested in by Developers. Part V includes:
 Chapter 21, ImageFinder Dos Version; and
 Chapter 22, API.
Chapter 23 provides the Reference Manual, which describes all Commands, Filters, and Parameters.

Chapter 24 is the readme.txt file, which quickly describes software requirements, installation, and the image recognition process using the ImageFinder.

If you just want to learn what the ImageFinder can do, use Trail 1. If you just want to have a quick test of what the ImageFinder can do, use Trail 2.

Trail 1. Evaluating ImageFinder Technology

Trail 1 includes reading Part 1 (Chapters 3, 4, 5) in any order. You will learn what image recognition can do.
Trail 2. Learning the Basic Operations
Trail 2 includes Part III (Chapters 6, 7, 8, 9, 10, and 11). You will learn the basic operations of the ImageFinder.
Trail 3. Learning the ImageFinder.
Trail 3 includes Part III, Part IV, and Part V. You will learn more details of operations and the multi-layered approach to image recognition.
Return