Charge-based prediction circuits for focal plane image compression

This work presents the design of a computational charge-based circuit to be part of a focal plane compression chip. The image compression scheme pursued is predictive coding. The proposed circuit computes the prediction error at every pixel. It carries out the computations by integrating the photocurrents of the pixels in a small neighborhood. The prediction weights for every pixel can be changed by changing the switching timing of the circuit making possible the use of adaptive prediction algorithms. The circuit is compact and can be integrated at the pixel level.