SERVICES >Forms Mapping & Forms Recognition

Automated Forms Recognition

Built around an image decomposition engine our Automated Forms Recognition module is a proprietary SWolution. This module enables recognition of scanned image types and further narrows down to define only the areas that contain the desired data in the images – we call the latter part of the operation “zoning”. The image decomposition engine actually decomposes the image first into black and white dots and then group them into letters, words etc. based on our home grown algorithm.

To define a new form type, the domain experts from our team analyze the form in detail, its various peculiarities and general layout and they make a draft design of an algorithm / logic to be applied for that form to get recognized. This is verified and reviewed and then we implement the algorithm in VBScript language. Our runtime engine executes the logic implemented in VBScript against each image being processed and verifies the result to get the zones in the image if the form got recognized.

To develop a rule set for a form type, we have developed our own integrated development environment (IDE) wherein the image will be displayed on one side and on the other side the VBScript functions can be written. It also facilitates the developer to debug the functions, step into, set break points, watch the intermediate values and so on and so forth. Once the rule set is found to be working as expected, the script can be saved for later use in different workflows.

For complex forms like HCFA1500 which contains something like 60-70 zones that need to be captured, we get a recognition rates well above 98%. For simple forms or letters, the requirement would be only to capture data from 3 to 5 zones; in such cases the recognition rate is as high as 100%. Our research team works continuously on such forms to get better recognition since each additional zone being recognized automatically will end up in huge savings in terms of data-entry & operator time.