The extra symbol you see in the right most column is showing that top left segment for clarity (that was the intention, at least : ) - let's call it the 'filter' for now. Ok so the second row has been sorted left and right by taking the digits that have a top left segment in their symbol representation and putting them on the left, those that don't have it go on the right. I can pull the connection to segment F (top left) and still tell which number is being displayed but, if I pull any more then some numbers can be confused. I've a 7 segment display running on a breadboard in front of me. Great in theory, but which three segments would you use to distinguish all the numbers from 0 to 9? The lower bound of sensors, if it continued to follow close to a 50/50 'binary' split would be log 27 = 2.8.Ĭome to think of it - now I'm thinking that maybe BCD uses similar concepts itself? Of course your photodiode would have to be appropriately masked for that area. with a little work figuring it out you could have less than 7 sensors.įind the primary distinguishing 'element' - for instance, maybe it could be that both right hand side units a lit (0,1,3,4,7,8) - then branch a tree depending on that element.

With 7 bits a single BCD character should afford you 128 unique symbols - considering there are only 10, there is a lot of redundancy. Step seven: output the most likely characterĪnother way is to implement a neural network and train it with characters. Step six: compare the bitmap with a collection of bitmaps representing chars Step six: normalize the bounding box e.g. Step five: draw bounding box around the char Step four: isolate separate characters (and/or words) Step three: apply morphological filters to skeletize the text, remove noise, speckles etc Step two: convert the image to black and white The latter can be done with a MEGA (I programmed a simple OCR long ago on a C64 in Basic so a MEGA must work)

Simple OCR recognizes one type of font in one size. High end OCR recognizes printed text in most fonts and sizes. OCR is not one algorithm, it is the collection of all image to text algorithms.īut there are differences, the very high end OCR recognizes written handwriting (banknotes, signature recognition etc)

Tesseract 7 segment display font