Συμπιεστική δειγματοληψία & συμπίεση εικόνας
SubjectΕπεξεργασία εικόνας -- Ψηφιακές τεχνικές ; Επεξεργασία σημάτων ; Ψηφιακά φίλτρα ; Αλγόριθμοι
The compression process constitutes the most usual method for storage and transmission of images in contemporary applications. Given this, the algorithms that exist for image compression can reduce the volume of data, making functionally efficient the systems that acquire information like for example high-resolution systems of image acquisition. Although several studies that concern image compression have been conducted, the essence of this issue remains unvaried. In fact, we transform an image in a appropriate basis and then code only the most significant of its coefficients. Therefore, our problem is reduced to the finding of a really good transformation. In the present dissertation we will refer to and study the aforementioned transformation. The image compression algorithms convert the high-resolution images into relatively small bit streams (while keeping the essential features intact), whilst at the same time the most significant elements of the image remain unchanged. The issue that emerges is whether there is some way to create data compression immediately and at the same time of data acquisition. The answer is absolutely positive and this is implemented by Compressive Sensing or Compressed Sampling (C.S.). In the present master thesis, we deal with this new relatively concept firstly theoretically and then our approach is extended with precision and accuracy to the mathematical background, which is required for its better and deeper understanding. After the above-mentioned approach we put forward some examples which refer to Compressed Sensing (C.S.) and its results. Finally, with the help of the appropriate software that we have created and with the use of a suitable compressed sensing algorithm we can observe clearly how compressive sensing works and what results produces as regards both the size of the initial and final image and of course its desired quality. The final results as well as the measurements are presented at the end of the present project for the judgment of the reader.