LDR Camera #7 : 98% Working

17 Oct 2019

Enter Your Project for a chance to win an Oscilloscope Grand Prize Package for the Most Creative Vision Thing Project!

Back to The Project14 homepage

Project14 Home

Monthly Themes

Monthly Theme Poll

The LDR camera is working! Well, almost. 98% = 63 of the 64 pixels are working. Just one is refusing to work properly which is not too bad as it has only just been finished constructing and wiring everything together. It is probably loose or broken connection somewhere, possible a broken LDR as I had to use quite a lot of force to fit the LDRs into the holes drilled into the wood for them. I will have to work out which pixel it is in the camera and try tracing the wiring. I'm a little bit reluctant to do too much messing and moving it about as there are a lot of wires and connectors involved, nearly 150 in total. The Dupont connectors only seem to make a really good connection the first time they are put together. Once they are disconnected I assume there has been a little bit of give in the springiness and subsequent insertions are just not as tight and with this many wires I need as much tightness as possible. The video below shows the LDR camera working and displaying the letter C.

The next step (apart from tentative explorations of the non-working pixel and/or permanently fixing the whole contraption to a piece of wood to improve robustness, is to somehow transfer the data to the BeagleBone AI. I am in two minds as to whether to try and control the LDR directly from the BeagleBone AI - which would be the best solution, or to somehow transfer the LDR camera data to the BeagleBone AI so that it can be processed.

I am going to try using an Artificial Neural Network to process the LDR camera data, to try and get it to recognise the 26 letters of the alphabet. This will be a C programme based on one I have used before, from a book called Adaptive Pattern Recognition and Neural Networks by Yoh-Han Pao. It is a pretty old book (1989) but I obtained this copy only a coupe of years ago so it might still be available somewhere on the internet. A full listing of the programme is given in the appendix, which I have typed in and then made some major revisions to, to improve the user interface, file handling and similar.

Here is the beginning of the programme as listed in the book. I'm not sure what the copyright situation is as it is so old.

The problem I will have with the programme is that is saves the ANN data set into a file and I do not know if the BeagleBone AI comes with a file system. I'm guessing not so I will have to amend the program to not use files. This might be beyond me in the time remaining but I will see what I can manage.

Dubbie