Nonlinear adaptive processing of block coded images

Implementation and measurement: A 32 x 32 array is fabricated using a 0 . 8 ~ SPDM CMOS process. Each pixel occupies 88 x 9 2 p * with a fill factor of -25% as shown in Fig. 3. The dimensions of the whole chip are 4196 x 4196pmz. During the experiment, a patterned laser light spot is incident upon the chip. The output current of each pixel is accessed sequentially using two shift registers for X and Y addressing. Signals are further amplified by an on-chip readout circuit. Fig. 4a is the original image, while Fig. 4a-c show the smoothed images, with an increasing degree of smoothing. The effects of edge enhancement, with and without smoothing, are illustrated in Fig. 4d and e, respectively. Fig. 4f shows the detected edges and Fig. 4g and h show the extracted horizontal and vertical edges, respectively. Introduction: Many well known image compression techniques such as JPEG [1] are block based. In these techniques, an image is partitioned into small blocks (typically 8 x 8) and each block is coded independently. However, at low bit rates the reconstructed images generally suffer from visually annoying artefacts due to very coarse quantisation. One such major artefact is the blocking effect, which appears as artificial boundaries between adjacent blocks. There are many techniques developed to reduce the blocking effect. Some use image filtering techniques [2, 31, and others formulate the blocking effect removal as an image restoration problem [5].