Two weeks ago I taught for the very first time the new course “Hands-On Evaluation of Image Sensors”. The course location was Barcelona, and the organization was done by CEI-Europe. Teaching a new course for the very first time is always a bit an adventure, because you do not know what the participants expect from it. But this time it was even trickier, a complete new course and for the very first time with measurement equipment in the class room. I organized 10 laptops, 10 cameras, 10 light sources, 10 power supplies, testcharts, lightboxes, etc., all needed to perform hands-on evaluation of image sensors and cameras. Fortunately the day before the course start I had a day off, so I had plenty of time to install the equipment in the class room and check out the hardware and software. To be prepared for any further hardware disaster while running the course, my daughter Kim was standby during the two days, she has more experience with hardware and a soldering iron than I have myself. But besides some minor issues, the hardware worked smoothly. The soldering iron could remain in the travel case.
Day 1 : after a short introduction, I went through a first exercise together with all the participants. The assignment was to proof that the noise in images is decreasing after averaging several images and show experimentally that the noise reduction is inverse proportional to the square root out of the number of images. By itself a simple exercise to get acquainted with the equipment, to get familiar with the measurement software and to get a first idea about the difference between temporal noise (non-correlated in the images) and fixed-pattern noise (correlated in the images).
After this first getting-acquainted exercise I put all the participants to work in groups of two. Their first assignment was to measure all fixed-pattern noise components of an unknown imager in an unknown camera by grabbing and analyzing images in dark. Once the images were stored on the hard-drive, a short piece of code needed to be developed to calculate all parameters from the images obtained. Some participants struggled a bit with writing the software, but after all I was surprised how quickly everyone was able to grab the images. At the moment all groups obtained their results, the theory behind the exercise was explained and the parameters measured/calculated were discussed and compared.
Next measurement assignment focused on obtaining the temporal noise parameters of the image sensor and the camera based on the same set of images obtained earlier. The same way of working was followed : first all groups worked separately and afterwards the results were discussed in a plenary session.
Day 2 : while the first day was focusing on doing measurements in dark, during the second day the lights were turned on. Again in two sessions the fixed-pattern noise components were measured/calculated and later all temporal components were evaluated. All these measurements were done quite quickly, because the code developed on the first day could be reused during the second day. Measurements with light on the sensors at different exposure times gave rise to the famous “Photon Transfer Curve”. All participants could experience how you can construct the PTC based on data obtained from multiple images. But during the theoretical session it was also explained how you can generate a PTC based on just three images, on only two images and even a single image. At the end of day 2 attention was paid on how to measure the MTF of a camera (based on a single image !) as well as on how to measure the spectral response of a camera (based on a single image !)
The overall feedback of the participants after the course was quite positive. Several reactions sounded like : “I have learned a lot !”. It is always good to hear that my “customers” are satisfied, but nevertheless after the first time running this course, I learned also a lot and will start to fine-tune the course for the next time. What is going to change ? Here is a list of action items I defined for myself :
– Some extra software functions/tools will be developed that can be used during the measurements/calculation so that the participants can focus more on the interpretation of the data,
– Extra images will be grabbed and stored on the laptops so that once the participants have the measurement code ready, it can be used for more different situations than the ones possible/available in the class room. For instance images generated by other type of cameras, or images generated at lower or higher temperatures,
– Updating and optimizing the course material, this is always necessary after the first edition of a course. Based on the questions and remarks from the participants during the course, I learned about the quality of my own sheets.
Although I did already during the evaluation of the course, I would like to thank the participants once again for the cooperation during the course, for their feedback afterwards and especially for attending this very first edition of the “Hands-On Evaluation” course. As far as I know this is a unique project in the world of imaging. No one else organizes classes with hands-on measurements and evaluation of commercially available cameras. It was also a very unique experience preparing this training and running it for the first time. The updating of the course towards the second edition will take quite a bit of work, but nevertheless I am looking forward to do it. The next “Hands-On Evaluation” course is scheduled for May 2011 in Copenhagen.
Albert, 05-12-2010.