赞
踩
色彩方面一直是短板,一直有两个疑惑,如何标定?出了问题,如何调试?今天主要研究了下标定的具体实现。
GitHub - QiuJueqin/color-correction-toolbox: Camera Color Correction Toolbox
代码关注截图3个文件即可,其他主要取色生成src和tar
有这么几个疑惑:
1.拟合训练的模型,具体的损失函数
2.拟合训练的过程如何保证不影响白平衡:[1 1 1]*CCM和白点tar的差异加到loss里面
CCM 标定的原理是,使用 sensor 抓拍到的 24 色卡场景下前 18 个色块的实际颜色信息和其期望值,计算 3x3 的 CCM 矩阵。输入颜色经 CCM 矩阵处理得到的颜色与其期望值差距越小,则 CCM 矩阵就越理想。
海思CCM矩阵:8bit 小数精度。bit 15是符号位,0 表示正数,1表示负数。典型的三组 CCM 为D50,TL84,A 三个光源下的 CCM。典型的五组 CCM为 10K,D65,D50,TL84,A 五个光源下的CCM。优先红、绿、蓝纯色;然后肤色块
采集步骤:(1)标准 X-Rite 24 色卡,照度为 600Lux 均匀光源;(2)调整 AE 目标亮度,最亮灰阶(Block 19)的 G 分量亮度在饱和值的 0.8 倍左右
手动调试CCM步骤:
(1)优先确认颜色问题是由于AWB模块引入还是CCM引入
(2)过灯箱客观指标,确认是否需要重新标定
(3)最后再考虑手动调整CCM
a.一般都是针对单一颜色进行微调整(比如车牌天蓝,想要深蓝;人脸偏红或是偏黄绿),所以单一颜色都会有个当前RGB值,想要调整的目标RGB值,其实调整CCM的目的就是通过CCM使得当前RGB值趋于目标RGB值
b.针对a所述,调整RGB值,调整R:a00、a01、a02(RinR、GinR、BinR);调整G:a10、a11、a12(RinG、GinG、BinG);调整B:a20、a21、a22(RinB、GinB、BinB)。
CCM矩阵色彩校正的方法_一只特立独行的zhu..-CSDN博客_ccm调试
WDR模式下标定CCM,CCM 容易受到 DRC 的影响,容易造成颜色难以矫正:
(1)曝光比手动最大,同时也要调整亮度值,避免长帧过曝,采集长帧的 RAW 数据进行 CCM 的标定。标定过程中可以适当的降低饱和度,不能选择开启 autoGain 功能。
(2)适当减少 DRC 曲线对图像亮度的大幅度提升,这样 DRC 对颜色的改变会较弱。此时,图像的亮度会有所降低达不到想要的亮度,这时,可以用 gamma 对亮度进行适当的提升。这样联调 DRC 和 gamma 模块,可以让整体的颜色调节更准确一些。
(3)对于 WDR 模式,因为大多场景是混合光源场景,容易出现亮处颜色偏色,人脸颜色偏红等问题,除了可以降低饱和度值以外,还可以使用 CA 模块对这些区域适当的降低饱和度。
反gamma后图像和gamma 1.0存在些许差异:
(1)CSC模块
CCM模块代码:
https://github.com/WaterdropsKun/isppipeline_Python/tree/main/lecture14
Python实现ISP pipeline代码:
GitHub - WaterdropsKun/isppipeline_Python: isppipeline_Python
如有帮助,希望帮忙github帮忙点个star
- def degamma_hisi(data, clip_range, gamma_txt):
- # gamma degamma
- hisi_gamma_x = 1024-1
- hisi_gamma_y = 4096-1
-
- hisi_degamma_x = 256-1
- hisi_degamma_y = 1.0
-
- gamma_hisi_x1023_y4095 = []
- degamma_x255_y1 = []
-
- with open(gamma_txt, "r") as f:
- for i, line in enumerate(f.readlines()):
- line = line.split(',')
-
- gamma_hisi_x1023_y4095 = [float(x) for x in line]
- # for j, value in enumerate(line):
- # print(j, value)
-
- # x = np.arange(0, 1024+1, 1) # np.arange(start, end+step, step) [start, end] end/step+1
- # plt.plot(x, gamma_hisi_x1023_y4095)
- # plt.show()
-
- for i in range(hisi_degamma_x+1): # for i in range(0, hisi_degamma_x+1, 1):
-
- for j, value in enumerate(gamma_hisi_x1023_y4095):
- if (value / hisi_gamma_y * hisi_degamma_x) >= i:
- degamma_x255_y1.append(j/hisi_gamma_x)
- break
-
- # x = np.arange(0, hisi_degamma_x+1, 1)
- # plt.plot(x, degamma_x255_y1)
- # plt.show()
-
- # degamma
- data = np.clip(data, clip_range[0], clip_range[1])
- data = np.divide(data, clip_range[1])
-
- height = data.shape[0]
- weight = data.shape[1]
- channels = data.shape[2]
-
- for row in range(height): # 遍历高
- for col in range(weight): # 遍历宽
- pv0 = data[row, col, 0]
- pv1 = data[row, col, 1]
- pv2 = data[row, col, 2]
-
- data[row, col, 0] = degamma_x255_y1[int(pv0*255)]
- data[row, col, 1] = degamma_x255_y1[int(pv1*255)]
- data[row, col, 2] = degamma_x255_y1[int(pv2*255)]
-
- data_show = data.copy()
- data_show = np.clip(data_show * clip_range[1], clip_range[0], clip_range[1])
- # gbr = rgb[...,[2,0,1]]
- # data_show = data_show[..., ::-1]
- data_show = data_show[..., [2,1,0]]
- cv2.imshow("data", data_show.astype(np.uint8))
- cv2.waitKey(0)
-
- return np.clip(data * clip_range[1], clip_range[0], clip_range[1])
-
-
- def gamma_hisi(data, clip_range, gamma_txt):
- # gamma degamma
- hisi_gamma_x = 1024-1
- hisi_gamma_y = 4096-1
-
- hisi_degamma_x = 256-1
- hisi_degamma_y = 1.0
-
- gamma_hisi_x1023_y4095 = []
- degamma_x255_y1 = []
-
- with open(gamma_txt, "r") as f:
- for i, line in enumerate(f.readlines()):
- line = line.split(',')
-
- gamma_hisi_x1023_y4095 = [float(x) for x in line]
- # for j, value in enumerate(line):
- # print(j, value)
-
- # x = np.arange(0, 1024+1, 1) # np.arange(start, end+step, step) [start, end] end/step+1
- # plt.plot(x, gamma_hisi_x1023_y4095)
- # plt.show()
-
- for i in range(hisi_degamma_x+1): # for i in range(0, hisi_degamma_x+1, 1):
-
- for j, value in enumerate(gamma_hisi_x1023_y4095):
- if (value / hisi_gamma_y * hisi_degamma_x) >= i:
- degamma_x255_y1.append(j/hisi_gamma_x)
- break
-
- # x = np.arange(0, hisi_degamma_x+1, 1)
- # plt.plot(x, degamma_x255_y1)
- # plt.show()
-
- # gamma
- data = np.clip(data, clip_range[0], clip_range[1])
- data = np.divide(data, clip_range[1])
-
- height = data.shape[0]
- weight = data.shape[1]
- channels = data.shape[2]
-
- for row in range(height): # 遍历高
- for col in range(weight): # 遍历宽
- pv0 = data[row, col, 0]
- pv1 = data[row, col, 1]
- pv2 = data[row, col, 2]
-
- data[row, col, 0] = gamma_hisi_x1023_y4095[int(pv0*1023)] / 4095.0
- data[row, col, 1] = gamma_hisi_x1023_y4095[int(pv1*1023)] / 4095.0
- data[row, col, 2] = gamma_hisi_x1023_y4095[int(pv2*1023)] / 4095.0
-
- data_show = data.copy()
- data_show = np.clip(data_show * clip_range[1], clip_range[0], clip_range[1])
- # gbr = rgb[...,[2,0,1]]
- # data_show = data_show[..., ::-1]
- data_show = data_show[..., [2,1,0]]
- cv2.imshow("data", data_show.astype(np.uint8))
- cv2.waitKey(0)
-
- return np.clip(data * clip_range[1], clip_range[0], clip_range[1])
-
-
- def CCM_convert(data, CCM, color_space="srgb", clip_range=[0, 255]):
- # CCM工作在线性RGB因此需要先进行degamma
- if (color_space == "srgb"):
- data = color.degamma_srgb(data, clip_range)
- data = np.float32(data)
- data = np.divide(data, clip_range[1]) # 归一化
- elif (color_space == "hisi"):
- data = degamma_hisi(data, clip_range, "./gamma_hisi_int.txt")
- data = np.float32(data)
- data = np.divide(data, clip_range[1]) # 归一化
-
- # matrix multiplication
- output = np.empty(np.shape(data), dtype=np.float32)
- output[:, :, 0] = data[:, :, 0] * CCM[0,0] + data[:, :, 1] * CCM[0,1] + data[:, :, 2] * CCM[0,2]
- output[:, :, 1] = data[:, :, 0] * CCM[1,0] + data[:, :, 1] * CCM[1,1] + data[:, :, 2] * CCM[1,2]
- output[:, :, 2] = data[:, :, 0] * CCM[2,0] + data[:, :, 1] * CCM[2,1] + data[:, :, 2] * CCM[2,2]
-
- # gamma
- if (color_space == "srgb"):
- output = output*clip_range[1]
- output = color.gamma_srgb(output, clip_range)
- elif (color_space == "hisi"):
- output = output*clip_range[1]
- output = gamma_hisi(output, clip_range, "./gamma_hisi_int.txt")
- return output
-
-
- if __name__ == "__main__":
- # CCM = np.array([
- # [1.507812, -0.546875, 0.039062],
- # [-0.226562, 1.085938, 0.140625],
- # [-0.062500, -0.648438, 1.718750],
- # ])
- CCM = np.array([
- [1.0, 0.0, 0.0],
- [0.0, 1.0, 0.0],
- [0.0, 0.0, 1.0],
- ])
-
- maxvalue = 255
- # image = plt.imread('kodim19.png')
- image = plt.imread('test02.png')
- if (np.max(image) <= 1):
- image = image * maxvalue
-
- new_image = CCM_convert(image, CCM, color_space="hisi", clip_range=[0, maxvalue])
- color.rgb_show(image / 255)
- color.rgb_show(new_image / 255)
Copyright © 2003-2013 www.wpsshop.cn 版权所有,并保留所有权利。