Graduate Project Proposals

1. Petrol Filling Station :(Source, Freelancer.com)

We are an independent Petroleum company about embark on an automation program for our third-tier filling stations. At present these stations – typically located in more high-tech developing countries are fully manual in operation. In general terms we wish to improve our quality of service, productivity and overall profitability. You are invited to tender for the supply of a computer-based monitoring and control equipment for use as part of the filling station system (as described below).
The specific aim of this project is to produce a trial system, to be used to evaluate the effectiveness of the (proposed) automated techniques. The major drivers in the system are both technical and operational. Moreover, some important constraints must also be taken into account when producing your proposal.  Further details can be found in this document [DOC]

1. Using CAPTCHA as typing helper

Handwritten documents are very hard to recognize by computer. No image analysis technique can recognize handwritten letters with 100% accuracy. To overcome this problem, human can help us recognizing handwritten documents free of charge. An example of handwritten document is shown here


The handwritten document is manually cut into small pieces marked with marker. Each piece will be an input to CAPTCHA. The document after marking will be as follow

The grid is entered to image processing system that will extract each piece and save each piece with an ID to the database.
The database is part of the website database. The website must have many visitors each day. From the website control page, you can implement a page for uploading the grid document and save it to database. Each piece of data has pieceID, imageLocation, digitalValue, digitized?, etc.

When the user login, CAPTCHA is displayed containing one of the pieces together with regular CAPTCHA as shown in the figure

In the figure "Cookie" word is handwritten word, and to the right the regular captcha word. The system must make sure that the user entered at least 1 extra digit beside the regular captcha word. The system can't recognize the word cookie, so we are not sure that the user entered the right word or not. The user will enter something like "cookieadmips", the system will cut this string into two pieces, and extract the piece to the left then save it to the database. The system uses the piece to the right for the authentication purpose.

Simple percentage bar can be displayed to show how many pieces are digitized. After completing 100% of the document, the system collect back the words in the same order using the pieceID and display the result into web-based document.