采用Python实现快速YUV转RGB
2021/7/20 11:06:39
本文主要是介绍采用Python实现快速YUV转RGB,对大家解决编程问题具有一定的参考价值,需要的程序猿们随着小编来一起学习吧!
目标
采用Python脚本实现快速的YUV图像二进制(BIN)文件到sRGB-24bit图像的转换,并保存为PNG文件。
解决方法
一般来说,YUV转RGB的主要手段有三种:
- libYUV, from Google Chromium Project, Open Source.
- FFmpeg, the most popular tools to decode images and video across all web/pc platforms.(BiliBili也是用的这个库做的视频解码)
- OpenCV, open source project mainly providing all popular and stable algorithms in computer vision.
各自特点:
采用libYUV会更快(一般4倍),采用FFmpeg效果可能更贴近人眼视觉(?),但采用OpenCV最方便,因为libYUV需要从源码根据平台修改编译,并没有库的形式直接调用,FFmpeg虽然有库,但是对python不友好(可能是我个人对基于FFmpeg的库不了解?),opencv就很简单,二进制的Yuv数据读进来就能直接转sRGB,而且和Numpy无缝结合,极为便利。
Show me the code right now!
import numpy as np import cv2 import sys import os import platform import glob def Windows(): return platform.system() == "Windows" YUV_NV21 = 0 YUV_NV12 = 1 def DecodeYUV(fp, w, h, c, t): # file pointer, width, height, channel, YUV type if c == 3: size_ = w*h*3//2 else: size_ = w*h data = f_.read(size_) if c == 3: im_yuv = np.frombuffer(data, dtype='<'+str(size_)+'B').reshape([h*3//2, main_w]) else: im_yuv = np.frombuffer(data, dtype='<'+str(size_)+'B').reshape([h, w]) if c==1: return im_yuv im_rgb = None if t == YUV_NV21: im_rgb = cv2.cvtColor(im_yuv, cv2.COLOR_YUV2RGB_NV21) elif t == YUV_NV12: im_rgb = cv2.cvtColor(im_yuv, cv2.COLOR_YUV2RGB_NV12) else: print('not implemented yet!') assert False return im_rgb if __name__ == '__main__': assert len(sys.argv) == 2 print('================ CONVERT YUV 2 RGB FOR PRAGUE ONLY =============') # create dirs for output images dir_main = 'picMain' dir_aux = 'picAux' dir_bokeh = 'Bokeh' cmd_mkdir = '' if Windows(): cmd_mkdir = 'md ' else: cmd_mkdir = 'mkdir -p ' os.system(cmd_mkdir + dir_main) os.system(cmd_mkdir + dir_aux) os.system(cmd_mkdir + dir_bokeh) yuv_dir = sys.argv[1] print('YUV DIR : ' + yuv_dir) print('****** PROCESSING MAIN ******') # find main images main_w = 3264 main_h = 2448 files = glob.glob(yuv_dir + '/*_input*_3264x2448.yuv') for i in range(len(files)): print(files[i]) f_ = open(files[i], 'rb') im_ = DecodeYUV(f_, main_w, main_h, 3, YUV_NV12) f_.close() # save RGB images fn_start = '' if Windows(): fn_start = files[i].rfind('\\') else: fn_start = files[i].rfind('/') fn_ = dir_main + files[i][fn_start:] fn_ = fn_.replace('.yuv', '.png') cv2.imwrite(fn_ , im_) print('****** PROCESSING AUX ******') # find aux images aux_w = 1600 aux_h = 1200 files = glob.glob(yuv_dir + '/*_1600x1200.yuv') for i in range(len(files)): print(files[i]) f_ = open(files[i], 'rb') im_ = DecodeYUV(f_, aux_w, aux_h, 1, YUV_NV12) f_.close() # save RGB images fn_start = '' if Windows(): fn_start = files[i].rfind('\\') else: fn_start = files[i].rfind('/') fn_ = dir_aux + files[i][fn_start:] fn_ = fn_.replace('.yuv', '.png') cv2.imwrite(fn_ , im_) print('****** PROCESSING BOKEH ******') # find aux images bokeh_w = main_w bokeh_h = main_h files = glob.glob(yuv_dir + '/*_out_*.yuv') focus_file = open('focus.txt', 'wt') for i in range(len(files)): print(files[i]) f_ = open(files[i], 'rb') im_ = DecodeYUV(f_, bokeh_w, bokeh_h, 3, YUV_NV12) f_.close() # save RGB images fn_start = '' if Windows(): fn_start = files[i].rfind('\\') else: fn_start = files[i].rfind('/') fn_ = dir_bokeh + files[i][fn_start:] fn_ = fn_.replace('.yuv', '.png') cv2.imwrite(fn_ , im_) # save the focus information from filename beg_id = files[i].rfind('Focus') end_id = files[i].rfind('_levle') focus_str = files[i][beg_id + 5:end_id] if i==0: focus_file.write(focus_str) else: focus_file.write('\n' + focus_str) focus_file.close() print('================================================================')
这篇关于采用Python实现快速YUV转RGB的文章就介绍到这儿,希望我们推荐的文章对大家有所帮助,也希望大家多多支持为之网!
- 2024-11-21Python编程基础教程
- 2024-11-20Python编程基础与实践
- 2024-11-20Python编程基础与高级应用
- 2024-11-19Python 基础编程教程
- 2024-11-19Python基础入门教程
- 2024-11-17在FastAPI项目中添加一个生产级别的数据库——本地环境搭建指南
- 2024-11-16`PyMuPDF4LLM`:提取PDF数据的神器
- 2024-11-16四种数据科学Web界面框架快速对比:Rio、Reflex、Streamlit和Plotly Dash
- 2024-11-14获取参数学习:Python编程入门教程
- 2024-11-14Python编程基础入门