Voucher Image Character Recognition via Sectional Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for character recognition on voucher images, such as receipts, often lead to false recognition of important items like telephone numbers, dates, and total amounts due to inappropriate categorization, making it difficult to accurately extract these details.
Innovation Solution
An information processing device that uses a learnt model to divide voucher images into sectional images and determine their types based on a dataset associating images with type information, employing machine learning techniques like neural networks to accurately identify and extract important character strings.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If OCR-based character recognition is performed on voucher images, then character strings can be extracted, but false recognition occurs and categorization accuracy deteriorates
Solution Approach 1:
The voucher image is divided into multiple sectional images corresponding to different regions (header, body, footer). Each sectional image is independently categorized using a learnt model to determine its type (e.g., telephone number, date, total amount). This segmentation approach allows the system to focus on specific regions with their characteristic formats, significantly reducing false recognition while maintaining high extraction efficiency.
2Speed
If simple character recognition methods are used, then processing speed is maintained, but recognition accuracy of important items deteriorates
Solution Approach 1:
A learnt model is pre-trained using a dataset containing numerous voucher images with manually annotated correct answers for each sectional image. This preliminary learning phase enables the model to acquire categorization knowledge before actual processing. During operation, the pre-trained model quickly categorizes new sectional images without requiring complex real-time analysis, thus maintaining high processing speed while achieving accurate recognition of important items.
3Device complexity
If traditional categorization methods are applied, then processing simplicity is maintained, but false categorization of items increases
Solution Approach 1:
A learnt model acts as an intermediary between the raw sectional image and the final categorization result. The model receives the sectional image as input and outputs the categorized type (e.g., telephone number, date). This intermediary layer processes the image data through learned patterns and relationships, enabling reliable categorization while keeping the overall system structure simple and easy to implement.
Data Source
AI summary
The information processing device includes a storage section storing a learnt model, a reception section, and a processor. The learnt model is obtained by mechanically learning the relationship between a sectional image obtained by dividing a voucher image and a type of a character string included in the sectional image based on a data set in which the sectional image is associated with type information indicating the type. The reception section receives an input of the voucher image to be subjected to a recognition process. The processing section generates a sectional image by dividing the voucher image received as an input and determines a type of the generated sectional image based on the learnt model.