This week was filled with a mixture of happiness and woes, brainstorming and formulating new techniques, and reading. Below is a detailed description of the week:

Monday

We started off our week still having problems with the Harris.py algorithm. Our main goal was to have an algorithm that will match identical corner points of two images. With no success, we had an emergency meeting with our mentor, Dr. David Crandall, who used a series of mathematical desk checks and coding diagnosis to solve most of the errors in our already generated code. It still didn’t show a report on image details, but the code was able to show key corner points in two images and match them.

We still worked on obtaining better results from our code.

Tuesday

We still worked on the algorithm, trying out more and improvised functions which will help produce better results. Our matching algorithm needed to be accurate before we include any report generating algorithm on identical features.

Wednesday

As usual, Gerald and I continued working on the algorithm. We still were facing many stumbling blocks with our prototype code, but still pushed to perfect our results. We also had a faculty meeting at 3 o’clock in the afternoon to report on the progress of our project. We also brought to light the bug which we identified in the Harris.py example in the “Programming Computer Vision” book, and were informed that if the Harric Corrner Detection technique doesn’t work, we were allowed to use different techniques. Dr. Crandall also sent a sample copy of his new Computer Vision paper to the SROC students with the purpose of expanding our scope in the area and receiving feedback from us regarding the book.

Thursday

We spent majority of the day reading the first part of Dr. Crandall’s 32-paged paper and working on our algorithm. The paper addressed the limitations of computer vision, and we made notes along the way. We also worked feverishly on our code – watching YouTube videos and trying to collate every piece of information we could get in respect to image processing and image matching.

Friday

The week was over, but the worry of our final goal not being met haunted me. I spent the early hours of the day just observing our algorithm; tweaking with thresholds and desc ranges. I also gave feedback to Dr. Crandall regarding his paper. The remainder of the day was spent brainstorming with Gerald on our project, and testing out newer techniques, which at the end produced little or no result.