TPJ.OCR 1.0.1

TPJ.OCR

Simple OCR library to make OCRing images easy!

Thanks

TPJ.OCR uses a number of other open source packages so special thanks to:

• Tesseract
• Charlesw Tesseract Wrapper
• IRON Drawing

How to use TPJ.OCR

First copy the tessdata, x64, and x86 folders from the example app TPJ.TestOCR from github, set each of the files from within those folders to copy if newer. This is the tesseract engine that will perform the OCR.

You must convert the image you wish to OCR into a AnyBitmap file, if you are loading it from disk then this is simple done like so:

AnyBitmap.FromFile(filePath);

AnyBitmap has a number of loading methods will should cover any way you need.

Next call one of the OCR extension method on the AnyBitmap object bitmap.OCR() - See examples.

Content Type

OCR is very hard and it can often produce incorrect characters such as 1 where it should be an I or a 3 where it should be an S. If you know the data you are OCR is going to be only digits or only alphabetic we can change the common issues with the correct characters.

There are three content types you can pick from:

1. Digits - Only digits (0 - 9) characters
2. Alphabetic - Only alphebtic (Aa - Zz) characters
3. Any - This is the default any means no changes are made from the result of the OCR

OCRArea, OCRRectangle, Location

There are three classes within TPJ.OCR OCRArea, OCRRectangle, Location.

Location

This contents the start and end pixel for the rectangle area that will be OCRed. If you have the image you are going to OCR you can simply open the image in paint and note down the X and Y for the top left of the rectangle and the bottom right.

OCRRectangle

This holds the X and Y location and a method to create the rectangle object based off the X and Y location.

OCRArea

This holds your rectangle location and allows you to set the content type (Digits, Alphabetic, Any) and set a name for the area, the name is useful when you are OCRing a single image for mutiple areas as the result will contain the name given in the OCRArea object plus the result of the OCR allowing you to confidently pick out the result for the area.

Examples

Check out the examples in github

Full image OCR

AnyBitmap.FromFile(@"./ExampleImage.png").OCR();

Single Area Digits Only

AnyBitmap.FromFile(@"./ExampleImage.png")
    .OCR(ContentType.Digits, 
    new OCRRectangle()
    {
        X = new Location()
        {
            StartPixel = 460,
            EndPixel = 700,
        },
        Y = new Location()
        {
            StartPixel = 220,
            EndPixel = 320,
        }
    });

Multiple Areas

var result = AnyBitmap.FromFile(@"./ExampleImage.png")
    .OCR(new List<OCRArea>() 
    {
        new OCRArea()
        {
            Name = "Name",
            ContentType = ContentType.Alphabetic,
            Location = new OCRRectangle()
            {
                X = new Location()
                {
                    StartPixel = 300,
                    EndPixel = 700,
                },
                Y = new Location()
                {
                    StartPixel = 40,
                    EndPixel = 140,
                }
            }
        },
        new OCRArea()
        {
            Name = "Payroll Name",
            ContentType = ContentType.Alphabetic,
            Location = new OCRRectangle()
            {
                X = new Location()
                {
                    StartPixel = 1350,
                    EndPixel = 1700,
                },
                Y = new Location()
                {
                    StartPixel = 40,
                    EndPixel = 140,
                }
            }
        },
        new OCRArea()
        {
            Name = "HRMS No",
            ContentType = ContentType.Digits,
            Location = new OCRRectangle()
            {
                X = new Location()
                {
                    StartPixel = 460,
                    EndPixel = 700,
                },
                Y = new Location()
                {
                    StartPixel = 220,
                    EndPixel = 320,
                }
            }
        },
        new OCRArea()
        {
            Name = "Pay Date",
            ContentType = ContentType.Any,
            Location = new OCRRectangle()
            {
                X = new Location()
                {
                    StartPixel = 1170,
                    EndPixel = 1710,
                },
                Y = new Location()
                {
                    StartPixel = 220,
                    EndPixel = 320,
                }
            }
        }
    });