FUNCTION REPRESENTATION OF AN IMAGE BY MEANS OF CSRBF - AN APPLICATION FOR RESOLUTION CONVERSION

摘要

As the need to use various devices increases today (e.g. computer monitor, digital camera, cellphone), the production of images at different resolutions is needed. In this paper, we propose CSRBF (Compactly-Supported Radial Basis Functions) image format to store CSRBF coefficients. CSRBF is popular for scattered data interpolation. Since this image format has function representation of an image, resolution conversion can be easily performed with high precision. Because resampling is easy once it performs function representation by means of CSRBF. Furthermore, in an environment with restriction such as storage or network, we propose CSRBF using wavelet transform preprocessing (WTP) to reduce points for image compression. This technique can reconstruct a high precision image with less image degradation than the conventional techniques. We show the superiority by data reduction rate comparison of CSRBF interpolation and conventional techniques.