原创不易 转载请注明出处和作者
转载地址:https://www.cnblogs.com/luoyinjie/p/7219319.html
关于写这篇文章的原因:
本人也是摸索了很长时间才弄懂其中的原理,里面涉及的知识点太多了, 比如色彩空间, Gamma, 什么的, 里面还会涉及到很多的协议, 比如 BT601 BT709, BT2020,RP177 等, 一不小心就会写错, 然后色彩可能就不准确了, 可能偏白 偏黑或者偏色
自己也踩过很多坑, 然后把这些经验写出来, 防止大家踩重复的坑, 如果文章有不够严谨的地方, 请及时指出。
YUV与RGB互转的公式有很多, 不同的色彩空间的转换公式是不一样的, 不同色彩空间的转换公式可以看我写的这篇文章: https://www.jianshu.com/p/db1caf7f2920
除了色彩空间, 还需要注意YUV的范围, YUV 有很多种表现形式, 比如:
YUV : YUV是一种模拟型号, Y∈ [0,1] U,V∈[-0.5,0.5]
YCbCr :也叫YCC, YCbCr 是数字信号, 它包含两种形式, 分别为TV range 和 full range, TV range 主要是广播电视采用的标准, full range 主要是pc 端采用的标准, 所以full range 有时也叫 pc range
TV range 的各个分量的范围为: YUV Y~[16,235] Cb~[16-240] Cr~[16-240]
full range 的各个分量的范围均为: 0-255
我们平时接触到的绝大多数都是 YCbCr (tv range) , ffmpeg 解码出来的数据绝大多数也是这个, 虽然ffmpeg 里面将它的格式描述成YUV420P , 实际上它是YCbCr tv range
YUV转tv range: Y' = 219.0*Y + 16 ; Cb = U * 224.0 + 128; Cr = V * 224.0 + 128;
关于为什么要将YUV量化为tv range 16-235 ?
以下是*摘抄的一段, 意思是tv range是为了解决滤波(模数转换)后的过冲现象,
Y′ values are conventionally shifted and scaled to the range [16, 235] (referred to as studio swing or "TV levels") rather than using the full range of [0, 255] (referred to as full swing or "PC levels"). This practice was standardized in SMPTE-125M in order to accommodate signal overshoots ("ringing") due to filtering. The value 235 accommodates a maximal black-to-white overshoot of 255 − 235 = 20, or 20 / (235 − 16) = 9.1%, which is slightly larger than the theoretical maximal overshoot (Gibbs phenomenon) of about 8.9% of the maximal step. The toe-room is smaller, allowing only 16 / 219 = 7.3% overshoot, which is less than the theoretical maximal overshoot of 8.9%. This is why 16 is added to Y′ and why the Y′ coefficients in the basic transform sum to 220 instead of 255.[9] U and V values, which may be positive or negative, are summed with 128 to make them always positive, giving a studio range of 16–240 for U and V. (These ranges are important in video editing and production, since using the wrong range will result either in an image with "clipped" blacks and whites, or a low-contrast image.)
关于如何判断像素格式是否为tv range (16-235)?
在完全黑画面的时候打印出图像的Y数据, 如果Y=16左右 说明YCbCr 为tv range ,如果Y=0左右 说明YCbCr为 full range
以下是我推导的 BT601 与 BT2020 的公式, 包括各种形式, 比如 浮点 整形 等等。
一、 公式:基于BT.601-6 (NTSC 色域)
BT601 UV(CbCr)的坐标图(量化后): (横坐标为u,纵坐标为v,左下角为原点)
通过坐标图我们可以看到UV(YUV六面体投影到UV坐标系)是一个旋转了一定角度的八边形, U越大蓝色越蓝,V越大,红色越红。
以下具体为各种转换公式(该转换公式基于BT601 ,NTSC色域 )
1.小数形式, YUV ( U∈[-0.5-0.5] , R,G,B∈[0,1] )
R = Y + 1.4075 * V;
G = Y - 0.3455 * U - 0.7169*V;
B = Y + 1.779 * U;
Y = 0.299*R + 0.587*G + 0.114*B;
U = (B-Y)/1.772;
V = (R-Y)/1.402;
或写为:
Y = 0.299*R + 0.587*G + 0.114*B;
U = -0.169*R - 0.331*G + 0.5 *B ;
V = 0.5 *R - 0.419*G - 0.081*B;
2.整数形式(减少计算量)未量化 R,G,B~[0,255] U,V~[-128,128]
R= Y + ((360 * (V - 128))>>8) ;
G= Y - (( ( 88 * (U - 128) + 184 * (V - 128)) )>>8) ;
B= Y +((455 * (U - 128))>>8) ;
Y = (77*R + 150*G + 29*B)>>8;
U = ((-44*R - 87*G + 131*B)>>8) + 128;
V = ((131*R - 110*G - 21*B)>>8) + 128 ;
3. 量化为 tv range 后的公式( Y~(16,235) U/V ~(16,240) )
[Y,U,V,1]T = M[R,G,B,1]T 其中 M =
[ 0.2568, 0.5041, 0.0979, 16
-0.1479, -0.2896, 0.4375, 128
0.4375, -0.3666, -0.0709, 128,
0, 0, 0, 1 ]
[R,G,B,1]T = M[Y,U,V,1]T M =
1.1644 0 1.6019 -223.5521
1.1644 -0.3928 -0.8163 136.1381
1.1644 2.0253 0 -278.0291
0.0000 0.0000 0.0000 1.0000
4 tv range的公式写成整数的形式(减小计算量) ( Y~(16,235) U/V ~(16,240) )
yuv --> rgb
R = (298*Y + 411 * V - 57344)>>8
G = (298*Y - 101* U - 211* V+ 34739)>>8
B = (298*Y + 519* U- 71117)>>8
rgb --> yuv
Y= ( 66*R + 129*G + 25*B)>>8 + 16
U= (-38*R - 74*G + 112*B)>>8 +128
V= (112*R - 94*G - 18*B)>>8 + 128
5. YUV量化 与 非量化 互转
tvrange 转 fullrange
Y=(Y'-16 )*255/219 ;
U=(U'-128)*128/112;
V=(V'-128)*128/112;
full range 转 tv range U~(-128-127) -----> U~(16-240)
Y' = ((219*Y)>>8) + 16;
U' = ((219*U)>>8) + 128;
V' = ((219*V)>>8) + 128;
7. 矩阵形式(BT601):
YUV
[Y,U,V]T = M[R,G,B]T 其中 M = 0.299 , 0.587, 0.114, -0.169, - 0.331, 0.5, 0.5, - 0.419 - 0.081
[R,G,B]T = M[Y,U,V]T 其中 M = 1 0 1.4017 1 -0.3437 -0.7142 1 1.7722 0
YCbCr tv range
[Y,U,V,1]T = M[R,G,B,1]T其中 M = [ 0.2568, 0.5041, 0.0979, 16 -0.1479, -0.2896, 0.4375, 128 0.4375, -0.3666, -0.0709, 128, 0, 0, 0, 1 ]
[R,G,B,1]T = M[Y,U,V,1]T M = 1.1644 0 1.6019 -223.5521 1.1644 -0.3928 -0.8163 136.1381 1.1644 2.0253 0 -278.0291 0.0000 0.0000 0.0000 1.0000
tvrange的公式写成整数形式
[Y,U,V,1]T = (M[R,G,B,1]T)>>8 其中 M = 66, 129, 25, 4096, -38, -74, 112, 32768, 112, -94, -18, 32768, 0, 0, 0, 256
[R,G,B,1]T = (M[Y,U,V,1]T)>>8 M = 298, 0, 410, -57229, 298, -101, -209, 34851, 298, 518, 0, -71175, 0, 0, 0, 256
二、. Rec2020 (BT2020) 下的YUV与RGB转换公式 (写成矩阵形式)
BT2020 UV 的坐标图(量化后为CbCr): (横坐标为Cb,纵坐标为Cr,左下角为原点)
1. BT2020 文档上的公式
即:
Y = 0.2627*R + 0.6780*G + 0.0593*B;
U = -0.1396*R - 0.3604*G + 0.5*B;
V = 0.5*R - 0.4598*G -0.0402*B;
矩阵形式
YUV RGB互转公式
[Y,U,V]T = M[R,G,B]T 其中 M = 0.2627 0.6780 0.0593 , -0.1396 -0.3604 0.5000, 0.5000 -0.4598 -0.0402
[R,G,B]T = M[Y,U,V]T 其中 M = 1.0000 -0.0000 1.4746 1.0000 -0.1645 -0.5713 1.0000 1.8814 -0.0001
YCbCr(tv range) RGB互转公式
[Y,U,V,1]T = M[R,G,B,1]T 其中 M = 0.2256, 0.5823, 0.05093, 16, -0.1222, -0.3154, 0.4375, 128 , 0.4375, -0.4023, -0.0352, 128, 0,0,0,1
[R,G,B,1]T = M[Y,U,V,1]T M =1.1644, 0, 1.6853, -234.3559, 1.1644, -0.1881, -0.6529, 89.0206, 1.1646, 2.1501, 0.0000, -293.8542, 0.0000, 0.0000, 0.0000, 1.0000
tv range 互转公式写成整数形式
[Y,U,V,1]T = (M[R,G,B,1]T)>>8 其中 M =
58, 149, 13, 4096,
-31,-81, 112, 32768,
112, -103, -9, 32768,
0, 0, 0, 256
[R,G,B,1]T = (M[Y,U,V,1]T)>>8 M =
298, 0, 431, -59995,
298, -48, -167, 22789,
298, 550, 0, -75227,
0, 0, 0, 256