Events
Superresolution - classical and novel concepts
00:00 07-01-2007
The keyword superresolution appears in at least 1341 publications. In general, it refers to system and signal processing techniques used to increase the resolution dictated by the finite bandwidth. In this talk I’ll present a classification of superresolution systems and techniques. I’ll identify three classes of superresolutions, two quite common and one new. With the first class, superresolution is achieved with systems for which the bandwidth does not provide the correct tool to asses the resolution, therefore by appropriate post processing, supperresolution can be achieved. With the second class, superresolution is achieved by multichannel measurements together with appropriate digital processing algorithms. The newest superresoltion approach uses very recent concepts from the theory of compressed sensing and sparsity. I’ll present results from my research using each of the three approaches. For the first class I’ll demonstrate complete reconstructions form undersampled (in Nyquist sense) digital holograms. For the second class I’ll present some studies on the applications of digital suprerresolution algorithms on images obtained with vibrating camera and 3D images obtained with integrated imaging technique. For the third class I’ll present a novel imaging system that uses the compressed sensing concept. Superresolution is achieved in sense that the reconstructed image has more pixels than the sensor has. The imaging system used for this purpose is completely not conventional in sense that it is not based on common traditional imaging elements.