DCPM差分预测编码

jehhdhdd

于 2022-06-05 14:36:04 发布

阅读量278

点赞数

文章标签：音视频

本文链接：https://blog.csdn.net/jehhdhdd/article/details/125131667

版权

DCPM差分预测编码

一编码原理

以本次实验为例，本次实验对图像的各个像素点进行压缩，并且采用左侧预测的方法，即每一个像素点的值输入后，与左边前一个像素点的预测值进行相减，随后经过量化与反量化得到差值，而后用来更新当前的预测值，并为下一个进来的值做准备，那么这个原理的实现过程中，误差的产生主要集中于量化与反量化之间产生的误差，采用不同的量化方式时误差也会不同。

二代码实现

# include<stdio.h>
#include<iostream>
using namespace std;
#pragma warning(disable:4996)
unsigned char* yuanshi = NULL;
unsigned char* chazhi = NULL;
unsigned char* yuce = NULL;
int main(int argc, char** argv) {
	int b = 2;//8比特量化b=2 4比特量化b=32 2比特量化b=128 1比特量化 b=256
	char c;
	char* yuvname = NULL;
	yuvname = argv[1];
	int W, H;
	W = atoi(argv[2]);
	H = atoi(argv[3]);
	char* yucename = NULL;
	char* chazhiname = NULL;
	yucename = argv[4];//预测图像
	chazhiname = argv[5];//差值图像
	yuanshi = (unsigned char*)malloc(W * H);
	chazhi = (unsigned char*)malloc(W * H);
	yuce = (unsigned char*)malloc(W * H);
	FILE* yuvfile;
	yuvfile = fopen(yuvname, "rb");
	FILE* yucefile;
	yucefile= fopen(yucename, "wb");
	FILE* chazhifile;
	chazhifile = fopen(chazhiname, "wb");
	fread(yuanshi, W * H, 1, yuvfile);
	int i, j;
	for (i = 0; i < H; i++) {
		for (j = 0; j < W; j++) {
			if (j == 0) {
				yuce[j + i * W] = yuanshi[j + i * W];
				chazhi[j + i * W] = 0;
			}
			else {
				c = (yuanshi[j + i * W] - yuce[j + i * W - 1]) / b;
				chazhi[j + i * W] =128+c;
				yuce[j + i * W] =yuce[j + i * W-1] + c* b;
				 
			}
		}
	}
	fwrite(chazhi, W * H, 1, chazhifile);
	fwrite(yuce, W * H, 1, yucefile);
	unsigned char* a = NULL;
	a= (unsigned char*)malloc(W * H/2);
	for (i = 0; i < W * H / 2; i++) {
		a[i] = 128;
	}
	fwrite(a, W * H / 2, 1, chazhifile);
	fwrite(a, W * H / 2, 1, yucefile);
	fclose(yuvfile);
	fclose(yucefile);
	fclose(chazhifile);
	free(a);
	free(yuce);
	free(chazhi);
	free(yuanshi);
	return 0;
}

输出了两幅图像，是经过解码以后的图像，另外一幅是差值图像，由于差值有正有负，在这里预处理把差值都加上128，得到插值图像。