Form Classification via Line-Art Alignment and Background Subtraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing technologies face challenges in accurately classifying forms with misregistered data and subtracting printed background form information, leading to poor performance in zonal OCR applications.
Innovation Solution
A method and system that utilize line-art junctions to generate fingerprint frequency descriptors for document type classification, align the image with candidate template models, and perform background subtraction using distance transforms to enhance OCR accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-generated harmful factors
If connected component filters are used to remove printed background form information, then background subtraction is achieved, but key character features (such as dots, dashes, etc.) and parts of broken up characters are wiped out
Solution Approach 1:
The patent segments the form image processing into distinct stages: template matching to identify document type, registration to align with template, and then selective background subtraction. This segmentation allows preserving character features by only removing background elements that do not overlap with registered character locations
Solution Approach 2:
The patent performs preliminary registration of the form image to the template model before background subtraction. This preliminary alignment establishes accurate correspondence between image features and template locations, enabling selective removal of background form information while preserving character data that aligns with expected positions
2Productivity
If template matching classification is performed on subsampled version of original image for fast processing, then processing speed is improved, but classification accuracy deteriorates
Solution Approach 1:
The patent applies partial action by performing template matching on a subsampled version of the image for initial classification, which is sufficient for most cases. This partial processing achieves fast classification without requiring full-resolution analysis, balancing speed and accuracy appropriately
3Manufacturing precision
If registration is performed using corresponding points derived by matching image blocks to special fiducial markings or image anchor templates, then alignment is achieved, but the process fails when data is misregistered with respect to data fields
Solution Approach 1:
The patent introduces template models as an intermediary reference that contains idealized representations of form layouts for different document types. Instead of directly matching image blocks to fiducial markings, the system first matches the entire form image to the appropriate template model, then uses this template-guided registration to align data fields, providing robustness against misregistration
Data Source
AI summary
A system and method to classify forms. An image representing a form of an unknown document type is received. The image includes line-art. Further, a plurality of template models corresponding to a plurality of different document types is received. The plurality of different document types is intended to include the correct document type of the unknown document. A subset of the plurality of template models are selected as candidate template models. The candidate template models include line-art junctions best matching line-art junctions of the received image. One of the candidate template models is selected as a best candidate template model. The best candidate template model includes horizontal and vertical lines best matching horizontal and vertical lines of the received image, respectively, aligned to the best candidate template model.