摘要 |
<p>PURPOSE:To reduce computing quantity by calculating a second predicted error which regulates a motion vector, and detecting the minimum predicted error which regulates the motion vector out of the second predicted errors calculated by an arithmetic means by a comparison means. CONSTITUTION:Memory 3 stores the information of a prediction target image in block unit, and memory 4 stores the information of a retrieval area in a reference image, and reads out the information according to the information from address generating parts 1, 2. The predicted error PE1 is found by a predicted error arithmetic part 5 by using L1 picture elements in a block, and the error PE1 is compared with a threshold value TH by a comparison part 6. When the error PE1 is smaller than the threshold value, the predicted error PE2 is found by the arithmetic part 5 by using L2 picture elements in the block. The error PE2 is compared with the minimum value HIM-PE by a comparison part 7, and when the error PE2 is smaller than the minimum value, the error PE2 is stored in a register 8 as the minimum value MIN-PE with the motion vector. The motion vector stored lastly is set as the optimum motion vector.</p> |