|
Abstract:
|
JPEG2000 , the international standard for still image compression , uses the wavelet transform . The filters used in JPEG2000 for transformation are the 9 /7 Daubechies filters and the 5 /3 Le Gall filters . At present , to achieve image mirroring or rotation for images , one has to convert them back to spatial domain and proceed . This thesis presents methods by which it can be implemented in the wavelet domain , without extra computational complexity . To do so , one manipulates the transform domain coefficients so that the condition of perfect reconstruction still holds true .
The thesis also discusses a method to achieve image interpolation , which would result in a higher resolution image from a lower resolution one . One manipulates the wavelet domain coefficients of the image for the same . For comparison purposes , one takes the low frequency component obtained after the wavelet transform of the image , which acts as the low resolution image , try to predict the rest of the components , and then reconstruct the original image . The reconstructed image is then compared with the original image . |