Wavelet Image Resizing in the Block DCT Space

In this paper, we present algorithms for resizing and transcoding of images in the transform domain. The approach is based on wavelet filtering (analysis/synthesis) along with downsampling/upsampling operation in the block DCT space. We use linear filtering in the block DCT domain to perform convolution. To obtain results equivalent to linear convolution, filtering is performed on the three adjacent blocks. To reduce the complexity, we perform sampling rate change and filtering operations in a single combined step. The proposed approach achieves the best performance in terms of quality compared to the DCT domain technique.