论文部分内容阅读
本文提出了多值制图象的高效率预测编码方法。它把各类连续变化的图象信号当作m阶Markov过程,每个象素的预测值取决于其邻近象素的非线性函数,而对预测误差的非线性量化,则取决于其邻近象素增量状态的类型。这种编码方法,充分利用图象信号在每个象素邻近区域的具体变化特点,逐个象素进行自适应非线性预测及量化。因此,提高了编码效能,特别是在图象的轮廓边沿或交织结构部位。计算机模拟表明,当传输信号的熵值为2bits/pel时,主观上难以发现图象损伤。本文的编码方法与其他高效率预测编码方法相比,不仅预测效率高,而且结构简单,适用于各种图象的高速实时传输。
This paper presents a multi-value system of high-efficiency prediction coding method. It treats all kinds of continuously varying image signals as m-order Markov processes. The prediction value of each pixel depends on the non-linear function of its neighboring pixels. The nonlinear quantization of prediction errors depends on the neighboring image The type of prime delta status. This coding method, make full use of the image signal in the neighborhood of each pixel of the specific changes in the characteristics of pixel-by-pixel adaptive non-linear prediction and quantization. Therefore, coding efficiency is improved, especially at the outline or interlaced parts of the image. Computer simulation shows that when the entropy of the transmitted signal is 2bits / pel, it is hard to find the image damage subjectively. Compared with other high-efficiency prediction coding methods, the coding method in this paper not only has high prediction efficiency but also has simple structure and is suitable for high-speed real-time transmission of various images.