The CSDK Engine is page oriented, the basic unit used in processing is typically the image of the page. Most functions require a page as an input parameter, but some API calls also operate on rectangular areas of the image (zones). The Image handling module enables the Engine to handle several pages at the same time. The Engine can only manage and process pages within its own memory space. The images must be loaded into the Engine by the Image file handling module, or the Scanning module. Possible image sources include files, scanning devices or the memory of the integrating application. The best resolution for B/W images is 300 or 400 dpi. The optimal recognition resolution for grayscale or color images is 150 to 300 dpi.

CSDK offers pre-recognition functions to apply image preprocessing procedures to images. These functions enhance the image quality, and yield more accurate auto-zoning (carried out by the Zone Handling Module) and recognition (carried out by the Recognition Module). Preprocessing can include any or all of the following steps:

• Inversion: automatic, programmed, or none

• Rotation: automatic, programmed, or none

• Deskewing: automatic, programmed, or none

• 3D DESKEW: automatic, programmed, or none

• Use of SET tools through the ActiveX Image Viewer Control (IVC)

In addition, image despeckling and resolution enhancements are performed internally to render better results, with or without the above transformations.

Additional tasks in this module include:

• User modification of the loaded images

• Detection and cutting book pages (when the image contains two book pages)

• Auto-detection of table location and structure for the Table recognition module

• Line information detection

• Auto-detection of border, box or frame location and styling