CS 615 - Coin Grading System

 

By: Richard Bassett

 

 

Goal of Project:

 

Develop an automated system that will be used to identify and grade valuable collectibles items such as rare coins providing consistent and repeatable results.

 

 

Rationale for Project:

 

Rare coins are presently graded by human hand and eye inspection that often produces varied, inconsistent and sometimes dubious results. A difference of a single grade can often mean thousands of dollars in the value of the asset. Judgment is suspect with subjectivity and great financial incentives entrenched in the process. 

 

 

CS 615 – 616 Desired System:

 

Develop a system that is capable of extracting feature and conditional (grade) information from scanned (front and reverse) GIF images of United States Business strike coins. Ideally the developed system should be able to properly identify the following from a scanned image:

 

-         Series (example: Lincoln Cent, Indian Cent, Jefferson Nickel, Roosevelt Dime….etc)

-         Denomination (cent, 5 cent, 10 cent, 25 cent, 50 cent, $1)

-         Year

-         Mintmark

-         Grade  (fair, about good, good, very good, fine, very fine, extremely fine, almost uncirculated & uncirculated)

 

The system would need to be trained with 100 – 200 sample coins of various denominations, series, grades and mintmarks. The client of this project will provide coins required for training and testing the system.

 

Project Tasks & Timeline Milestones:

 

  1. Gain Domain Knowledge – All team members should have a full understanding of the terminology, tasks required & technology required.
  2. Establish Image Standards – define which scanner (or camera), which software, what resolution will be used for capturing and storing images.
  3. Build Expert Visual Database – build a database that can be used for feature extraction with complete images and fractional images of the sample coins (train the system)
  4. Build Statistical Database (if required) – this database would be numeric representations of the trained images in #3
  5. Develop the software for identification and feature extraction – this is the heart of the system as it is the software to compare the scanned image to the visual (or statistical) database. This task should return series, denomination, year, mintmark & grade.

6.      Increase the trained database to include a larger sample set and other denominations by bringing the trained population up to 300 – 400 coins.

 

 

 

 

 

  1. Test a greater volume of user supplied coins against the expanded trained database, modifications and additions to the software and algorithms in #5 may be required at this point. Each time a user supplied coin is successfully identified and graded then it should be added to our visual database as accumulated knowledge.
  2. Develop an intuitive web-based GUI that allows users to submit a scanned image for submission to #5. This GUI should return the series, denomination, year, mintmark & grade to the user.
  3. Tie in obtained results to a values database - Push #5 & #6 further by setting up a values database (weekly values are available on the Internet) and obtain the value of a scanned coin when obtaining series, denomination, year, mintmark & grade.

 

 

Tasks 1 – 5 should be completed by Christmas break in December

Tasks 6 – 7 should be completed by March 1, 2003

Tasks 8 – 9 should be completed by the end of the Spring 2003 Semester

 

The entire project should be well documented in Word as each significant task is completed.

 

 

Previous Work Done in this area

 

The Spring 2002 Pervasive Computing class developed a system that took a scanned image and obtained statistical data on the scanned pixels in the image in terms of the Hue, Saturation & Brightness vectors

. The statistical data collected in step 2 allowed the team to determine which coins are similar to others in their trained database in terms of known grade. The results under certain conditions were quite good but some scanned images presented serious error conditions such as false positives, which is why additional software is needed in this area. Complete documentation from the endeavors of this project is available to the CS615 team that undertakes this new project.

 

 

 

About the Client:

 

Richard Bassett is the client for this project. Mr. Bassett is a Professor at Western CT State University and a DPS Student at Pace. Components of this project will be used in his dissertation, which is well underway. As such Professor Bassett expects to work closely with the team that elects this project.

 

Last Updated: Tuesday, May 28, 2002