Overview

The software is in the form of MATLAB source code. MATLAB is a well known scientific and technical computing language. You will need a copy of MATLAB to be able to run this code. A copy of the code and documentation is downloadable from here.

The analysis of the ISO targets is performed using slightly modified versions of Peter Burns MATLAB code released in the public domain, with thanks. This is available for download from ISO 12233 Slant Edge Analysis Tool sfrmat 2.0.

Elements of the software from sfrmat2 are authored by Peter Burns, (peter.burns@kodak.com) Copyright (c) 2003 International Imaging Industry Association, the remaining software is Copyright (c) 2006 Lester Wareham

My understanding is the core of the IMATEST system is also derived from Peter’s code.

The USAF 1951 chart was downloaded from a unidentified web source and modified to suit.

The software is designed to operate with two types of test target regions of interest, ISO 12233 slanted line targets and USAF 1951 targets. The targets elements are to be placed measurement regions. The ISO slanted line elements should be slanted at approximately 7.5 degrees relative to the imaging pixel coordinate system and the USAF elements aligned to the lens cylindrical coordinate elements.

Each measurement region is expected to have a USAF 1951 target which by its nature includes both orthogonal directions of test lines covering a resolution range of 16:1.

Each measurement region should also have two orthogonal ISO targets.

As delivered the configuration for the ”Std APS-C” target is for 11 measurement regions with the corner USAF regions tilted at +/-34 degrees. This agrees with the APS-C sensor aspect ratio and is close to full frame aspect ratio.

The “Small APS-C” target includes only ISO ROIs and is intended for testing longer lenses

In the case of the USAF measurements the user must select the desired line elements based on the target magnification. The software labels these in two pairs of approximately 10 and 30 lp/mm. The software will then measure and report the local normalized contrast ratio of these line elements.

The advantage of the USAF targets is it makes it possible to measure the ration of meridonal and sagittal MTF, a good indication of lens astigmatism.

The target used in the most of the published results was designed to be used at a magnification of up to 1:27 the active target measuring 0.6075M by 0.405M. This permitted easy measurements of lenses of up to 200mm in normal domestic rooms.

For lenses over 300mm a smaller ISO only target was used measuring 0.303M by 0.202M.

An image of each target is shown below:

The 11 regions are labeled A through K consisting of one dead centre, two short side edges, two intermediate long side ‘centers’, two long side edges and four corners. For result reporting and reduction of data the regions are grouped as in the below table.

Region	Location (Landscape)	Classification
A	Upper left corner	Corner
B	Upper edge middle	Edge
C	Upper Right Corner	Corner
D	Left edge middle	Edge
E	Left of centre	Centre
F	Centre	Centre
G	Right of centre	Centre
H	Right edge middle	Edge
I	Lower left corner	Corner
J	Lower edge middle	Edge
K	Lower right corner	Corner

The targets were constructed on a section of artist’s canvas board, this can be obtained from an art shop. All elements of the measurement regions were printed on a 600 dpi laser printer and stuck onto the board with double-sided tape.

The USAF and ISO targets must be printed aligned to the laser pixel system to maintain line edge acuity. So each item had a boarder to cut around defining the slant.

The below figure shows one of the regions with the Regions of Interest that the typical measurements are derived from.

And an additional example for a corner region showing the angled USAF ROIs:

The following methodology was used when doing the lens testing.

1) Set the target up square and true on a firm stand.

2) Light the target as evenly as possible.

3) Set the camera up on a sturdy tripod.

4) With the lens close to the centre of the target centre the tripod height and position so the centre of the lens and target are aligned. Keep checking that the sensor and target are in parallel planes (straight and true)

5) Position camera to frame the target with the lens at the test focal length, checking the camera is still straight and true relative to the target.

6) Set the camera to Av, and the lowest ISO and ensure the exposure is correct, adjust if required. Use RAW to avoid JPEG artifacts. Select the first f-stop. Ensure mirror lockup is enabled and use a remote release or release timer.

7) Take a frame of each f-stop.

8) Repeat 3 times throwing off the focus and refocussing.

9) Change to a new focal length or lens and repeat from 5).

10) When complete transfer the shots to your computer.

a) Convert the RAW files to TIF format using your standard capture sharpening. (I used the TRL Sharpening Toolkit PS Action "Capture Enhanced Edge Masked USM" 0.7, 0, 300 with a 65% layer blend). This should be just enough to correct for the sensors sampling apature.

b) It is probably most convenient to convert to sRGB and 8-bits although this is not a requirement. The TIF must not use compression.

11) Rename the TIFs to the filename and directory structure detailed in the software documentation and run the software.

12) Sit back, look at your results and see which lens is best.

Last Updated 05/06/2008

All material is supplied as is and without warranty, use at your own risk.

All opinions stated are the authors own.

All quoted information remains the copyright of the respective authors.

Any trade marks mentioned are the property of their owners.

Contact: (anti spam remove underscores)

Note this email gets spammed a lot so if you want to get your message through please prefix "Holding Tank" in the title

f_ w_ a_ r_ e_ 9_ 0_ @_l_i_n_e_o_n_e_._n_e_t