A Computational Approach to Edge Detection
IEEE Transactions on Pattern Analysis and Machine Intelligence
| Hi-index | 0.00 |
This paper proposes a vector quantization scheme that uses variable vector dimensions and improves the perceived quality of the reconstructed images by using an edge based cost function. The cost function used for encoding is modified to take into account the number of edge pixels in a particular encoding block. For a better subjective quality, the product of the fraction of the edge pixels and the fraction of the non-edge pixels should be minimized. This is included in the additive cost function of VQ and minimized as an unconstrained problem. This edge based VQ technique reduces edge artifacts like jagging, ringing and also VQ distortions that happen near object contours, thus improving the overall subjective quality of the reconstructed image.