image recognition program?
BlitzMax Forums/BlitzMax Beginners Area/image recognition program?
| ||
| Hello, I have searched the forums but cannot find what I am looking for, partly because I do not know what to search for. I am looking for a program that can recognize numbers and/or letters from an image, most likely a .jpg, and spit out these numbers in a .txt file so I can import them myself into Excel, or write some code to format it like excel, altho I hav yet to try that. Anyways, thanks for all your guy's help in the past, and here's to the future. Cheers! B |
| ||
| always the same "kind" of picture and the same font? Or completely different sources? |
| ||
| Like the letters on a sign-post by the road-side? Or from the page of a book? |
| ||
| i am an intern at a company, i guess you could call my official title title a finance analyst. I get jps that have the information i include in an excel spreadsheet to compare websites and their contracts we have with them, the marketable traffic of the different age groups, ethnic groups, as well as religious and additional categories. I include all these and make a graph to show the roi (return on investment) based on these figures and show the CFO and VP and such where the best places to invest are, and where to direct sales. I get jpgs, much like a screenshot of some of these figures and I have to key in all the numbers by hand, which tend to be 100 sets of individual figures. the font is usually the same. Times New Roman, at about I would say 12 or 14 size font, but it varies sometimes. I know that some scanners have a scan to file option in which you can scan in a text document and it recognizes the numbers and letters. But I do not want to print every picture and then scan it back it. although that would be easier that what I am doing now. Sorry I should have included all this information in the original post. Thanks! Cheers B |
| ||
| My tesseract module does OCR, but your images may not be of a high enough resolution for it to understand the text. Over 100 dpi works well. But you never know. |
| ||
| But you do not have a chance to get the original datas? I ask, because... ...if the screenshot is a screenshot of a web-page, you might have the chance to grap the html-content of the page instead a screenshot of it. Do you have any influence on how this screeenshot is made? Could it be also a PDF? Are you only interested in the datas inside the spreadsheet?. How many spreadsheets you would have to scan a day? Are they identic in size, back color, rows, headings, etc...This makes a difference how to proceed the scanning (more manual or more automatic). Can you send a link to a sample of one of those spreadsheets, jpgs? (of course without any relation to any customers, etc...) to find out the quality of the jpg comression. |
   |