CheckProcessing actions

Supported image formats

TThis library supports single page TIF and JPEG images with a resolution of 200 dpi to 300 dpi. If the image dpi is not within this range, recognition may be poor or may not occur. If your checks have an incorrect dpi, such as 96 dpi, which is common for a photo that is taken by a camera, you can use the ImageConvert action SetImageDPIByWidth to correct the dpi of the image by specifying the correct physical width of the check. For example, if the check should be 5.5 inches wide: SetImageDPIByWidth("5.5","300","True"). It is required that the dpi be within the 200-300 range and that the pixels / dpi reflects the correct physical size of the check.

US checks

The image must be black and white (1 bpp) or grayscale (8 bpp). Color is not supported for US checks. If you have color US checks, these checks must be converted before calling recognition..

Brazilian checks

Checks can be black and white (1 bpp), 8-bit grayscale (8 bpp) or 24-bit color (24 bpp) images. The Brazilian engine uses the color information when performing recognition, but color is not required.

Other international countries

For other countries, checks can be black and white (1 bpp), 8-bit grayscale (8 bpp) or 24-bit color (24 bpp) images. However, the color information is not used. The engine will automatically convert the image to black and white before performing recognition. If 24-bit color or 8-bit gray scale images are provided to these engines, these images are automatically binarized in memory before the recognition, leaving the original image unchanged.

If you want, you can use actions to binarize images can before processing so you have a black and white image of the check that is saved within the batch.

Check Borders

The check image must be the size of the check itself. Any borders or white space around the check image must be cropped making the image the same size as the physical check or recognition might not be correct or might not occur. The image enhancement ruleset can be used to deskew and remove image borders.

Example Ruleset

The following shows a specific example ruleset and functions that are configured to process checks originating from the US The page is first converted to black and white, and then run through the check processing action.

Ruleset Detect Checks

-Function Process Check

--ConvertImgageToBlackAndWhiteOCR_A("tio")

--ProcessCheck("USA", "false", "None")

It is strongly recommended that the document "Best Practices for optimal text recognition in IBM Datacap" is reviewed to understand how to get optimal results from recognition.

The Recognition actions cannot be used in the same application as the US check recognition. Check whether recognition for non-US countries can be used in the same application as the Recognition actions.