

We are talking about SSOCR, a tiny program that allows you to recognize digits of a seven segment display.

All regular algorithms will fail during the identification of the numbers, however, there's an easy way to identify such numbers with another tool. Pitifully, as you can see, the recognized characters were wrong recognized. In some application, you will need to use OCR to recognize automatically this number, however as you may know the recognition of such characters as the shown on this article, with regular OCR algorithms or with tools like tesseract, you won't be able to recognize the numbers correctly: An additional 8th LED is sometimes used within the same package thus allowing the indication of a decimal point, (DP) when two or more 7-segment displays are connected together to display numbers greater than ten. Each of the seven LEDs is called a segment because when illuminated the segment forms part of a numerical digit (both Decimal and Hex) to be displayed.

I have not trained my tesseract for LCD font.The 7-segment display (seven segment display), consists of seven LEDs arranged in a rectangular shape. Training tesseract for your font is not difficult and it will give your ocr very high level accuracy. For more accuracy you need to train tesseract for LCD font. The code may fail sometimes but tuning parameters of canny and appropriate dialation may improve results. To filter the images as clean as possible and then feeds it to Tesseract. The code uses opencv image filtering techniques Tried to perform OCR on noisy and dirty images (for eg. Tesseract is an open-sourced OCR which is capable of reading text from papers, pdfs and other clean formats. The satements and parameters of some funtions will changed according to the texture,Ĭolor, lighting effect and visibility of text of image. This is a tesseract based OCR to read from seven segment display.
