| In the last decade, we have witnessed an exciting boost in the performance of cell-phones. While their computational power is still far from that of a desktop computer, smart-phones are an interesting research platform for their portability, full programmability, and the fact that they are equipped with relatively high-quality cameras.;I will first exploit this potential to allow users to decode a barcode by simply taking its picture. My image-based barcode reader is able to extract all the information embedded in the graylevel values of the pixels---as opposed to requiring to work in the simplified, binary domain---with a high computational efficiency. My algorithm, which I also implemented on a Nokia N95 smart-phone achieving real-time performance, outperforms all published competitors.;Then, I will move to computational photography and propose the first complete pipeline to perform metering for stack-based, high-dynamic-range imaging. Implemented on a Nokia N900 smart-phone, my algorithm selects the number of pictures, and the relative exposure times, required for a specific scene; it generally requires the capture of smaller sets of pictures than competing methods while producing final images of higher quality.;Finally I will present my solution for avoiding the typical artifacts that occur in stack-based, high-dynamic-range imaging when the subject is not completely static. For such situations, and contrary to previously published methods, my approach is capable of locally maximizing the amount of dynamic range that is being captured while creating an artifact-free image. |
