Image Processing Apparatus Selective Region OCR
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing systems face inefficiencies in processing and displaying results, particularly when performing optical character recognition (OCR) on entire images, which can be time-consuming and unnecessary for tasks like file naming and meta-information extraction.
Innovation Solution
The system includes a processor that acquires an image, extracts regions with specific attributes, determines the most similar registered document based on positional information, selects a processing target region, performs character recognition on that region, and displays the text data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If OCR processing is performed on the entire image, then complete text recognition is achieved, but processing time increases significantly
Solution Approach 1:
The patent divides the image into multiple regions based on predetermined attributes (such as header, body, footer sections). Instead of performing OCR on the entire image, the system selectively applies character recognition processing only to specific regions that contain relevant information, thereby reducing processing time while maintaining recognition completeness for necessary text elements.
Solution Approach 2:
The patent applies different processing strategies to different regions of the image. High-priority regions with important information receive full OCR processing, while lower-priority regions use simplified or skipped processing. This local differentiation optimizes the balance between recognition accuracy and processing speed by allocating computational resources efficiently across different parts of the image.
2Loss of information
If character recognition is performed on all regions, then all text data is extracted, but processing efficiency decreases
Solution Approach 1:
The patent extracts and identifies specific regions with predetermined attributes (such as regions containing file names, dates, or other meta-information) before applying character recognition. By taking out only the necessary regions for processing, the system achieves complete extraction of required text data while avoiding the computational overhead of processing the entire image, thus improving processing efficiency.
3Measurement precision
If the system determines document type using OCR on entire image, then accurate document classification is achieved, but time consumption increases
Solution Approach 1:
The patent performs preliminary region extraction and identification based on positional information and predetermined attributes before conducting character recognition. By pre-processing the image to identify and segment relevant regions, the system prepares the data structure in advance, allowing for accurate document type determination using only the extracted regions rather than the entire image, thereby reducing time consumption while maintaining classification accuracy.
Data Source
AI summary
An apparatus acquires an image by reading a document, extracts a plurality of regions having a predetermined attribute from the acquired image, determines information about a registered document most similar to the acquired image from among information about a plurality of registered documents stored in a storage unit with use of positional information about the extracted plurality of regions, selects a processing target region in the acquired image based on a position of a processing target region previously specified with respect to the determined information about the most similar registered document, performs character recognition processing on the selected processing target region, and displays text data obtained by the character recognition processing.