We are a small company specialising in CCTV (Closed Circuit Television).
We produce high quality test charts for evaluating CCTV camera basic parameters.
The test chart is produced by colour calibrated process, so the A3 format printed chart has very accurate details and colours.
So far, users of out test charts have been doing manual visual analysis when comparing cameras or making their decision which camera and lens, which compression produces better results.
Based on the image produced and exported by a tested IP CCTV camera we want to be able to determine a few key parameters using software, rather than leaving that to the user for visual judgment.
We want talented visual processing programers to tell us if and how they can evaluate the following parameters:
1. Resolution, using the five Siemens stars on the test chart. The resolution needs to be expressed in LP/PH (Line Pairs per Picture Height). We will provide the formula for that, once we know how the programmers think this need to be done. The resolution to be plotted on a graph, as MTF versus LP/PH on the horizontal axis.
2. Colour reproduction, based on the standard TV colour bars at the top off the chart, with their pure RGB co-ordinates, but also based on the Macbeth colour chart printed in the test chart itself. The colour deviation to be shown as a Delta E on a graph.
3. Compression and/or sampling artefacts via the checking of the smoothness of the curves both in the continuous colour transition bar, as well as the continuous transition Black-White-Black B/W bar at the bottom part of the test chart. This also needs to be shown as a plot.
4. Determine the Gamma of the camera by checking the Grey level of the reproduced 11 steps of grey and reading the average grey level of each patch (average because there will be pixels which are higher values than other pixels in in the same patch). This needs to be shown as a plot too, showing B-W versus Grey percentage on the horizontal axis.
5. Determine the Noise level in the image by analysing the 50% grey level patch in more details, i.e. using the variation of the grey pixels and calculating the deviation of the various levels. S/N should also be produced from it, by reading the lowest and highest luminance level throughout the test chart. I can provide a formula for that too.
Images for analysis will typically be provided in HD (1920x1080) or UHD (3840x12160), by the customer, who would export such an image from their camera looking at the actual test chart.
The original, uncompressed TIF image of the test chart will be provided to be used as a reference and comparison with the image of the camera under test.
We know that most of these analysis can be done in MatLab, and we prefer programmers to produce the analysis software in this format. We are currently running that latest version R2018B on our computers, and we can verify and/or comment the program when in draft stage.
Ultimately, we would like to produce a Windows "exe" program from such program which we may offer to the customers buying the test chart as a part of the chart itself, or, alternatively, we would like to produce a web-based camera image analysis, where the users would upload their camera image for comparison and producing plotted measurements.
Please refer to the two uploaded images with hand-written explanation on the required points.
Also, when offering your work, I would appreciate if you describe how do you intend to make this happen.
Any questions, very happy to respond to the serious bidders.
The main thing i would attribute to the programmer would certainly be Matlab and a subset (App) within Matlab called Computer Image Processing. So a perso. Must know a little bit about image processing, video analytics and pixels
I believe the Matlab can produce a C program from it, or an “exe” once it works the way it is supposed to
By the Packaging function