Just a short update today...

More image tests

Using both the C270 and the C920, I took James Newton's idea and downloaded some street sign images. I also created another test document with some street names in really big font sizes and two slightly different fonts.

After the usual image conversion using "graphicsmagick" the resulting TIFF images were processed by tesseract-ocr.

Although I tried different image resolutions with the C920, and I also used the original downloaded images along with the photographed versions, most of the OCR conversions failed.

I'll definitely have to do some additional research and testing here. Since I'm already using graphicsmagick for the image format conversion, I can also use this to perform some filtering and image enhancement.

If I can find a nice GUI for graphicsmagick or imagemagick that runs with Debian Linux, I'll try to use that to find a good combination of image processing steps/options that will produce better OCR results. Once a good combination has been found, I can use this through the command line options of one of the tools. If there is no good GUI for both apps, I can switch to Gimp or something similar in order to find a good combination of image processing steps.

I'll also take a closer look at image processing algorithms and see if there probably already is a good solution for this specific problem.

Preparing the program

Apart from the image and OCR tests, I am working on the structure of the main "TextEye" application. Once I'm satisfied with the basic flow, I'll start to convert that into actual code (hopefully starting tomorrow).

Setting up this application and getting it to do what's needed seems easier right now than getting the image preprocessing right. I should be able to get some working result here even if the image processing is still off.

More hardware

I'm currently waiting on some parts I've ordered for this project. Sadly one of the parts - the Pimoroni "pHat DAC" board for the Raspberry Pi - isn't in stock right now, not even in Pimoroni's own web shop, and this stops the rest of the order being send out to me.

I will try to at least get the rest of the parts I don't currently have by placing an additional order. Hopefully this will allow me to focus on the physical side of things later this week or maybe next week.

Once the parts are halfway complete I can go about wiring the prototype, connecting the software to the hardware where necessary, and take some measures to develop a 3D design for an enclosure.