Encoder

摘要

An average quantization error value of each of I, P, and B pictures in an encoded unit of processing is calculated as an actually measured value having a large variation. An average quantization error value of each of I, P, and B pictures in an uncoded unit of processing is set as a target value having a small variation. In the encoding of the uncoded unit of processing, a result of the encoding of the encoded unit of processing is referenced and fed back thereto. By uniformly setting respective quantization errors of images and further by uniformly setting the respective qualities of the images, the image quality of a whole stream can be subjectively improved. Since the prefetch of the uncoded unit of processing is not needed, it is possible to perform real-time processing without any increase in the circuit scale.

主权项

1. An encoder for encoding an input image into an output data, comprising:
a quantization error calculation part configured to calculate a quantization error by comparing said input image with a decoded image obtained through decoding encoded data obtained through encoding said input image; a measured quantization error storage part configured to store measured quantization errors which are actually measured values of said quantization errors with respect to an encoded unit of processing which is already encoded; and a target quantization error setting part configured to set target quantization errors for images in an uncoded unit of processing which has not been encoded, the target quantization errors being target values of said quantization errors, the target quantization errors set by using a variation in said measured quantization errors of images in said encoded unit of processing in a manner that a variation in said measured quantization errors of images in said uncoded unit of processing is smaller than said variation in said measured quantization errors of images in said encoded unit of processing.