Many business processes dealing with image documents face the challenge of insufficient scanned document image quality for automated OCR. Virtual Image Processing component helps to optimize image quality for better recognition results using LeadTools technologies.
Documents with irregular color or patterned background have lower OCR success rates than ones with plain background. Adaptive thresholding, Color dropping, Color extraction and Thresholding filters and their combinations allow significant improvement of such images increasing their recognition rate.
Correct image alignment sometimes is a key feature to recognition success and is not always respected in scanned documents. This problem is solved using Deskew filter of VIP.
Color resolution filter increases/decreases bits per pixel, provides dither method, palette and color order parameters.
One of the most serious problems decreasing the quality of document OCR output is speckles (visual garbage) on original documents, especially faxes. Despeckle filter of VIP allows reducing the number of visual document flaws impeding the recognition process.
Dilation filter enlarges black objects in the image (omnidirectional, horizontal, vertical, diagonal dilation). This function is useful for making text easier to recognize, especially, when it is affected by copying or faxing.
Dot Remove filter removes dots of the specified size on 1 bit b/w document scans.
Flip filter flips the image.
Invert filter inverts all image colors.
Inverted Text filter inverts text inside black box and removes the box itself. It helps in extracting text, for example, from the table headers.
Line Remove filter removes horizontal or vertical lines on the scanned 1bit black and white document images. If line passes through text, the filter can be configured to remove the text or keep it. This feature is useful to improve the text recognition quality; for example, on the forms where text is printed over table borders or lines.
Resize filter allows changing output image size.
Reverse filter reverses the image (left to right) to produce a mirror image.
Rotate filter rotates the image to a fixed angle (90, 180, 270). This feature helps, when? for example, there is a batch of images resulting from documents incorrectly fed to a scanner.
Sharpen filter increases or decreases sharpness of the image.
Stretch intensity filter increases the image contrast by centering, maximizing, and proportioning the range of intensity values.