Document Boundary Detection Using Continuous Shadow Edges

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing image processing techniques for detecting the boundary between an object and a background member in a document are inaccurate when the shadow width changes due to variations in document thickness or illumination angle.

Innovation Solution

An image processing device that calculates an edge amount for each pixel in the shadow region between the object and background member, determines edge continuity, and detects the boundary position based on a preset number of consecutive edge pixels, irrespective of shadow width changes.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If conventional boundary detection methods are used, then detection speed is maintained, but detection accuracy deteriorates when shadow width changes due to document thickness or illumination angle variations

Engineering Contradiction:
Improveboundary detection accuracyVSAvoidadaptability to shadow width changes
Core Design Contradiction:
Measurement precisionVSAdaptability or versatility

Solution Approach 1:

The patent changes the detection parameter from fixed-distance pixel comparison to gradient-based edge detection. By calculating gradient magnitude and direction at each pixel location, the method adapts to varying shadow widths without requiring parameter adjustment, resolving the contradiction between detection accuracy and adaptability to shadow width changes

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent replaces the mechanical approach of fixed-distance threshold comparison with a field-based gradient calculation method. This substitution allows the detection algorithm to naturally adapt to continuous variations in shadow width caused by document thickness or illumination angle, improving both accuracy and versatility

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

2Measurement precision

If the detection method accounts for shadow width variations, then boundary detection accuracy improves, but processing complexity increases

Engineering Contradiction:
Improveboundary detection accuracyVSAvoidprocessing complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The gradient calculation automatically adapts to local image characteristics without requiring external parameter adjustment or complex preprocessing. Each pixel's gradient is computed independently based on its local neighborhood, allowing the method to handle shadow width variations inherently while maintaining relatively simple processing

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The patent segments the image into individual pixel evaluations, where each pixel's gradient is calculated independently. This segmentation allows parallel processing and avoids the need for complex global analysis, reducing processing complexity while maintaining high detection accuracy across varying shadow widths

Inventive Principle:
Principle #1Segmentation

Data Source

PatentUS20260057528A1Image processing device, image forming apparatus, image processing method, and non-transitory computer-readable medium
Publication Date: 2026.02.26 RICOH CO LTD
  • US20260057528A1 patent drawing
  • US20260057528A1 patent drawing
  • US20260057528A1 patent drawing

AI summary

An image processing device includes circuitry configured to calculate an edge amount for the target pixel in a shadow region in image data; determine whether the target pixel corresponds to an edge, based on the edge amount; determine whether the edge is a part of a continuous edge based on the edge amount, in response to a determination that the target pixel corresponds to the edge; count an edge continuity number for the target pixel; count an edge continuity number for a previous target pixel selected immediately before the target pixel and determine whether the edge continuity number for the previous target pixel is greater than a preset number in response to a determination that the target pixel does not correspond to the edge; and detect a boundary position based on the edge amount of each of the target pixels forming the continuous edge up to the previous target pixel.