Character String Detection Using Color Clustering

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing character detection techniques face challenges in accurately and efficiently detecting characters in images with complex backgrounds, particularly in languages like Japanese and Chinese, due to high processing times and low detection accuracy, especially in real-time video processing.

Innovation Solution

A character string detection device that uses clustering to group pixels by similar colors, generating a clipped image excluding non-similar color regions, allowing for targeted character detection within the clipped image, thereby reducing the influence of complex backgrounds and improving processing efficiency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If character recognition processing is applied to all character regions in an image, then detection accuracy is improved, but processing time increases significantly

Engineering Contradiction:

Improvecharacter detection accuracyVSAvoidprocessing time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The patent segments the image processing task by first extracting edge information to identify candidate character regions, then applying full character recognition processing only to these candidate regions rather than all regions in the image. This segmentation approach maintains detection accuracy while significantly reducing processing time by limiting intensive processing to relevant areas only.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent performs preliminary edge detection and candidate region extraction before applying character recognition processing. This preliminary action identifies potential character locations using computationally efficient edge operations, preparing a reduced set of candidate regions that require full recognition processing, thereby reducing overall processing time while maintaining accuracy.

Inventive Principle:

Principle #10Preliminary action

2Difficulty of detecting and measuring

If edge extraction is performed using Sobel or Laplacian filters to identify character regions, then character detection capability is improved, but detection accuracy decreases in images with complex backgrounds

Engineering Contradiction:

Improvecharacter region extraction capabilityVSAvoiddetection accuracy

Core Design Contradiction:

Difficulty of detecting and measuringVSMeasurement precision

Solution Approach 1:

The patent introduces an intermediary verification step between edge-based candidate region extraction and final character detection. The character recognition processing acts as an intermediary that verifies whether extracted candidate regions actually contain target characters, filtering out false positives from complex backgrounds while maintaining the ability to detect characters in various conditions.

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent implements a feedback mechanism where character recognition results are used to validate or reject candidate regions identified by edge extraction. This feedback loop allows the system to learn from detection outcomes and adjust candidate region selection, improving accuracy in complex backgrounds by eliminating regions that edge extraction incorrectly identified as characters.

Inventive Principle:

Principle #23Feedback

3Productivity

If conventional character detection methods are used in images with complex backgrounds, then processing can be performed, but detection accuracy decreases due to false extraction of background regions as character regions

Engineering Contradiction:

Improveprocessing capabilityVSAvoiddetection accuracy

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

The patent extracts and isolates the target character detection task from the complex background by using edge-based candidate region extraction followed by targeted character recognition. This extraction approach separates potential character regions from the complex background, allowing the system to maintain processing capability while improving accuracy by focusing recognition resources on relevant regions only.

Inventive Principle:

Principle #2Taking out (Extraction)

Data Source

PatentUS9129383B2Character string detection device, image processing device, character string detection method, control program and storage medium

Publication Date: 2015.09.08 OMRON CORP

AI summary

A character string detection device for detecting a character string including at least one character in an image has a clustering unit that defines at least one cluster including at least a fixed number of pixel groups having similar colors based on color information included in each pixel configuring the image, a clipping unit that partitions the image into a region of the pixel groups belonging to the cluster defined by the clustering unit and a region of other pixel groups, and generating a clipped image excluding the region of the other pixel groups from the image, and a character string detection unit that detects each character in a detection target character string to be detected so as to detect the detection target character string in the clipped image generated by the clipping unit.