AutoVISION Quick Start Guide December 2020 Precautions for Correct Use Direction of parameter adjustment To read correctly, click on character selection on the right side of the screen. The OCR algorithm works in 2 steps: Segment characters → Recognize characters. Some keys for success are described below. To check the detail for each parameter, see HELP. a. Segment characters i. Adjust min/max character size according to actual size in AutoVISION image view. To detect the characeter size, mouse over to characters and check the pixel coordinate shown in bottom part of AutoVISION. ii. Set [Scaling factor] as “Character width” x “Scaling factor” = 30 pixel. b.
Recognize characters Adjust Confidence below 0.8 (80%) to increase read rate. It is not uncommon that the same character has different shapes. Even at 50% it will not tend to misread, but will read the correct character in the face of more variation You can increase the recognition rate by adding characters which are miss-classified into the Font Library. Click on the character which should be added to library. The “Add sample” pop-up will appear. Type in correct character and it will be added to the sample set. The read confidence for each character can also be seen as a value of 0 to 1 when hovering over a character and has a color associated with the read confidence.
The colors are defined as below in relationship to match score: -Green: 81 – 100% -Orange: 51 – 80% -Red: 0 – 50%