vmunoz82 / pymjpeg Goto Github PK
View Code? Open in Web Editor NEWPure Python and an alternative C extension to perform full MJPEG decoding from MOV/MP4 files (2012)
Pure Python and an alternative C extension to perform full MJPEG decoding from MOV/MP4 files (2012)
This example shows a rather complex program in Python, it is a JPEG decoder that is able to decode (M)JPEG Baseline complain files. The image to be decoded could be in any chroma sub-sampling configuration, for example 4:4:4, 4:2:2, 4:2:0, or any custom one such that each plane is block quantity is a power of 2. Apart from plain JPEG files, this program can demux a .MOV (or .MP4) video file, and if it found a JPEG streaming on it, it dumps and decode every frame. The program has to be invoked as follows: python decode.py <image.jpg|video.mov> The procedure to this type of image processing usually require a lot of processing, for decoding a JPEG file consist of the following steps: -setup Huffman tables. -setup the quantization matrix. -decode a 8x8 block using Huffman codes and RLE. -apply a 2D Inverse Discrete Cosine Transformation. -process the chroma sub-sampling. -convert from the YCbCr color space to the RGB color space. Huffman decoding, and IDCT transformation make the JPEG decoding very Python unfriendly due to the speed of the Python bytecode execution, moreover, the YCbCr to RGB and the blit of the final image is very time consuming when applied to each pixel of an image. Inside this example I provide several examples to demonstrate Python performance optimization techniques. All these techniques are fine tuned with a very carefully profiling of code execution, and were tested decoding a video of 1180 frames of 320x240 pixels each (ftp://ftp.hp.com/pub/information_storage/software/video/MakeUp.mov) on a Core i5 CPU. The resumed results are shown in the following tables: CPython 2.7 2155.45 seconds. (1.0 speed factor) PyPy 1.7 (JIT) 198.85 seconds. (10.8 speed factor) CPython 2.7 modules C Huffman and IDCT modules. 393.75 seconds. (5.5 speed factor) Refactoring of Jpeg_decoder class for Cython and Huffman and IDCT C code 4.29 seconds. (502.4 speed factor) These experiment results shows that Python code could perform at outstanding speeds when it is properly profiled, and the CPU intensive critical parts are refactored to a lower level and higher performance implementations. In this case we decode a video containing 40 seconds of image in ~4.3 seconds, enough for realtime player, and ~500 times the speed of the pure Python implementation. Directory description: pure: pure Python version. common: C version of IDCT, YCbCr -> RGB, and Huffman algorithms. mod: CPython Huffman, IDCT and YCbCr C modules. mod/linx64_p27: Python 2.7 (Linux x86_64) binary modules mod/win32_p27: Python 2.7 (Windows 32 bits) binary modules fast: Jpeg_decoder class (jpegdec.py replacement) refoctored for Cython and integrated with native C version of Huffman, IDCT and YCbCr->RGB code. fast/linx64_p27: Python 2.7 (Linux x64) binary modules fast/win32_p27: Python 2.7 (Windows 32 bits) binary modules fast/win32_p32: Python 3.2 (Windows 32 bits) binary modules
A declarative, efficient, and flexible JavaScript library for building user interfaces.
๐ Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. ๐๐๐
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google โค๏ธ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.