| 摘要 |
<p>PURPOSE:To reduce the number of the encoded bits of the whole, and to improve encoding efficiency by predictive-encoding at first the right and lower picture element of every block of MXN picture elements of a picture, and next, predictive-encoding a point other than said picture element. CONSTITUTION:The picture divided into the blocks of MXN picture elements is inputted from an input terminal 1, and one picture portion is written in a memory 2. The picture element in the right and lower part in the prescribed block is predicted by a prediction circuit 7 according to the picture element which is adjacent to the block and was encoded before, and when its prediction error is large, the prediction error is quantized by a quantization circuit 4, and is encoded by an encoding circuit 9. The prediction precision of the picture element other than that in the right and lower part in the block is high and the prediction error is small, because both sides are encoded before. Accordingly, the number of quantized bits can be reduced, and besides, the number of bits whose prediction error is small enough and which need not be encoded increases as well, and the encoding efficiency is improved. Besides, since quantization is performed only when the prediction error is larger than a threshold, the encoding efficiency is improved.</p> |
