Week2 – Progress Report

During the last week I’ve investigated some image preprocessing algorithms used in Pattern Recognition problems. For the final paper I am finally more interested in extracting information from images rather than classifying them. Then I am focusing the research on feature extractors rather than classifiers.

The goal of the paper will be analyzing the performances (in terms of precision, computational complexity, velocity and noise rejection) of different features extractors – combined with the adequate preprocessing filters – to discern objects shapes. I will start with simple primitives, to gain confidence, then I’ll try with key’s profiles. Moreover, according to the time constraints and the skills acquired, I’ll try to interpret also the key’s section from a plain 2D image. Finally I’ll apply the simplest classification algorithm (i.e. instance-based ones like k-NN).

Going back to this week’s progress: I played around with WEKA with the segmentation dataset, I finished an online tutorial on basic Image Processing and I’ve start reading the Learning OpenCV book.

Finally I’ve planned my work for the next 10 weeks: Timeline.