Archived copy. This page was recovered from the Internet Archive snapshot of /display/SPR/Malformed TIFF images solution taken on 2013-07-16. The original wiki at wiki.opf-labs.org is being decommissioned.

Malformed TIFF images solution

Added by Swithun Crowe · last edited by Swithun Crowe · on Apr 18, 2012 (view change)

Title	Malformed TIFF images solution
Detailed description	A Python2 script takes files and directories as command line arguments. It attempts to open images using PIL (Python Imaging Library) and collects statistics on the images which can indicate how much black there is in the images. For an unmodified scan or photo, there shouldn't be any pure black. The output from the script can be opened in a spreadsheet application. Examining the TIFF files using hexdump, the black areas are caused by black pixels (0s), rather than un-terminated tags. In the sample, most TIFFs did not have black areas. Quite a few did, and a small number of TIFFs were nothing but 0s.
Solution Champion	Swithun Crowe
Corresponding Issue(s)	Black areas or pixels in TIFF images
Tool/code link	https://github.com/downloads/openplanets/SPRUCE/tiff_black_pixel_reporter.py also checked in as SPRUCE/tree/master/black_pixels
Tool Registry Link	Python Hexdump
Notes	The script is run as so: It will process any images below directory1, the files somefile.tif, another.tiff and any images below directory2. The output is in CSV: The first line shows an image which has no black pixels, so no percentage of black pixels. The second line shows an image which has a large number of black pixels, covering 95% of the image. The third line shows an image which is full of zeros. It couldn't be opened by PIL as an image, so the width and height couldn't be established.
Evaluation	Cause of problem determined, solution is quite straightforward Python script, uses imaging library High numbers of black pixels, strongly indicative of problem files Recurses through files and sub-directories Issues owner: This solves the problem, easy to use solution, investigation of cause of black pixels can now be made, useful for checking outputs of other digitisation projects. Jenny also had a similar problem previously, so is interested in this solution.