文件列表:

LICENSE (1291, 2017-07-17)
__init__.py (0, 2017-07-17)
jpegdir (0, 2017-07-17)
jpegdir\CHANGES (2644, 2017-07-17)
jpegdir\PORTABILITY (365, 2017-07-17)
jpegdir\SETUP (2601, 2017-07-17)
jpegdir\blkint.jpg (2899, 2017-07-17)
jpegdir\blkint.jpg.1 (16384, 2017-07-17)
jpegdir\blkint.jpg.2 (8192, 2017-07-17)
jpegdir\blkint.jpg.3 (8192, 2017-07-17)
jpegdir\chendct.c (8528, 2017-07-17)
jpegdir\codec.c (14854, 2017-07-17)
jpegdir\csize.h (2147, 2017-07-17)
jpegdir\dct.h (4348, 2017-07-17)
jpegdir\doc.ps (474103, 2017-07-17)
jpegdir\globals.h (7281, 2017-07-17)
jpegdir\huffman.c (16802, 2017-07-17)
jpegdir\io.c (30497, 2017-07-17)
jpegdir\jpeg.1 (6494, 2017-07-17)
jpegdir\jpeg.c (67996, 2017-07-17)
jpegdir\leedct.c (9325, 2017-07-17)
jpegdir\lexer.l (28678, 2017-07-17)
jpegdir\makefile (1641, 2017-07-17)
jpegdir\marker.c (18008, 2017-07-17)
jpegdir\marker.h (1783, 2017-07-17)
jpegdir\nonint.jpg (2922, 2017-07-17)
jpegdir\param.h (3946, 2017-07-17)
jpegdir\prototypes.h (5108, 2017-07-17)
jpegdir\stream.c (24441, 2017-07-17)
jpegdir\stream.h (1548, 2017-07-17)
jpegdir\system.h (3678, 2017-07-17)
jpegdir\tables.h (4706, 2017-07-17)
jpegdir\test.3stream (625, 2017-07-17)
jpegdir\test.huff (3789, 2017-07-17)
jpegdir\test.q (593, 2017-07-17)
jpegdir\transform.c (11621, 2017-07-17)
ljpeg.py (2966, 2017-07-17)
... ...

Usage: 0. Install flex with ``` apt-get install flex```. Building without flex will fail and pollute the codebase. 1. In your python project ``` git clone https://github.com/***lgo/ljpeg.git ``` Or if you are using git already: ``` git submodule add https://github.com/***lgo/ljpeg.git ``` 2. Produce the jpeg binary. ``` cd jpegdir; make ``` 3. In your python code: ``` from ljpeg import ljpeg x = ljpeg.read(path) ``` The loaded is a matrix of type uint16. Typically you want to convert that to float for subsequent processing. ``` x = ljpeg.read(path).astype('float') ``` The LJPEG format sometimes has wrong values for width and height (transposed). One has to read the correct values of width and height from the associating .ics file. Below is a sample snippet for this purpose: ``` W = None H = None # find the shape of image for l in open(ics, 'r'): l = l.strip().split(' ') if len(l) < 7: continue if l[0] == name: W = int(l[4]) H = int(l[2]) bps = int(l[6]) if bps != 12: logging.warn('BPS != 12: %s' % path) break assert W != None assert H != None if W != image.shape[1]: logging.warn('reshape: %s' % path) image = image.reshape((H, W)) ``` 4. Using ljpeg.py standalone: - Convert to TIFF (requires the .ics file in the same directory as LJPEG) ``` ./ljpeg.py cases/benigns/benign_01/case0029/C_0029_1.LEFT_CC.LJPEG output.tiff ``` - Convert to TIFF and verify that no information has been lost ``` ./ljpeg.py cases/benigns/benign_01/case0029/C_0029_1.LEFT_CC.LJPEG output.tiff --verify ``` - Convert to jpeg for visualization with down-sizing scale=0.3 (16-bit TIFF is not good for direct visualization) ``` ./ljpeg.py cases/benigns/benign_01/case0029/C_0029_1.LEFT_CC.LJPEG output.jpg --visual --scale 0.3 ``` Note that output file can be any format that's supported by OpenCV (which includes all common types). Most file formats only support 8-bit images, so directly saving into such file formats will cause problems. Add "--visual" to normalize color into 8-bit before saving to such file formats. The Stanford ljpeg code is in public domain and is therefore OK to be included here. I did minor modification to make the code compile under modern Linux.

近期下载者：

相关文件：

评论：[我要评论] [举报此文件]

收藏者：