In this letter, a modified dominant point detection algorithm is developed and applied for Interpolative Vector Quantization (IVQ) of images. The algorithm employs an optimal discontinuity detector to locate first order discontinuities which efficiently characterizes a scan line. Using the detected first order discontinuities as the initial guess, the final set of interpolating points is obtained using a Modified Split-&-Merge algorithm. Experimental results show that, under the same bit-rate, an average of 0.5 dB improvement is achieved by our proposed method when compared with the results obtained by the traditional IVQ in terms of Peak Signal-to-Noise Ratio (PSNR).
Scopus Subject Areas
- Signal Processing
- Computer Vision and Pattern Recognition
- Artificial Intelligence