An Efficient Fusion Based Up-sampling Technique for Restoration of Spatially Compressed Images

The various up-sampling techniques available in the literature produce blurring artifacts in the upsampled, high resolution images. In order to overcome this problem effectively, an image fusion based interpolation technique is proposed here to restore the high frequency information. The Discrete Cosine Transform interpolation technique preserves low frequency information whereas Discrete Sine Transform preserves high frequency information. Therefore, by fusing the DCT and DST based up-sampled images, more high frequency, relevant information of both the up-sampled images can be preserved in the restored, fused image. The restoration of high frequency information lessens the degree of blurring in the fused image and hence improves its objective and subjective quality. Experimental result shows the proposed method achieves a Peak Signal to Noise Ratio (PSNR) improvement up to 0.9947dB than DCT interpolation and 2.8186dB than bicubic interpolation at 4:1 compression ratio.