Image Reading Apparatus Dust Detection Using Line Sensor Sorting

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing image reading apparatuses face challenges in accurately detecting abnormal pixels, such as dust, without falsely identifying vertical lines as streaks, due to limitations in resolution and image processing techniques.

Innovation Solution

The apparatus employs a line sensor with staggered arrangement of red, green, and blue light receiving elements, sorting image data to ensure adjacent pixels have the same color, and using a dust detection circuit to identify abnormal pixels based on digital values, with optional expansion and correction processes to accurately detect and correct dust.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If binary data conversion and addition method is used to detect black streaks, then dust detection capability is improved, but false detection of vertical lines as black streaks occurs

Engineering Contradiction:
Improvedust detection accuracyVSAvoidfalse detection rate
Core Design Contradiction:
Measurement precisionVSReliability

Solution Approach 1:

The patent segments the dust detection process into multiple stages: binary conversion, line-by-line addition, threshold comparison, and false detection elimination through contextual analysis. By dividing the detection into discrete steps with intermediate validation, the system achieves accurate dust detection while eliminating false positives from vertical lines.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces an intermediary validation mechanism that acts as a mediator between the addition result and the final dust detection decision. This intermediary layer analyzes the spatial distribution and characteristics of detected streaks to distinguish between actual dust and vertical lines, preventing false detections while maintaining detection sensitivity.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Measurement precision

If multiple reading means are used to improve detection accuracy, then measurement precision is improved, but device complexity increases

Engineering Contradiction:
Improveabnormal pixel detection accuracyVSAvoidreading means configuration
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent makes the single line sensor multi-functional by enabling it to perform both image reading and dust detection functions. The same sensor data is processed through different algorithms: standard image processing for content recognition and specialized addition/threshold processing for dust detection. This eliminates the need for separate reading means while maintaining high detection accuracy.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Solution Approach 2:

The system uses its own reading data for self-diagnosis and dust detection purposes. The line sensor reads the image data, and this same data is then analyzed through the addition method to detect dust particles. The system serves its own detection needs without requiring external or additional reading devices, reducing overall system complexity.

Inventive Principle:
Principle #25Self-service

Applied Scientific Principles

This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.

Function Achieved in This Case

This approach enables precise detection and correction of dust without falsely identifying lines, improving image quality and reducing operational costs by eliminating the need for multiple reading means.

Implementation Method 1

a line sensor configured to photoelectrically convert the light reflected from the original, to thereby read the original

Methodology Applied
Scientific EffectPhotoelectric conversion: Photoelectric Effect

Data Source

PatentEP3796631B1Image reading apparatus
Publication Date: 2023.03.22 CANON KK
  • EP3796631B1 patent drawingFigure 1
  • EP3796631B1 patent drawingFigure 2
  • EP3796631B1 patent drawingFigure 3~4

AI summary

An original image is read as an aggregate of a plurality of pixels in which adjacent pixels have different colors (R, G, and B) in a main scanning direction and in a sub-scanning direction, and the read pixels of the respective colors are stored in a line memory in association with information on relative positions of the pixels with respect to another pixel. Then, the stored pixels are sorted so that pixels having the same color are adjacent to each other, and an abnormal pixel (dust) not present in the original image is detected based on the state of the sorted pixels. With this, the dust not present in the original image is detected without increasing the cost, and the dust is corrected without forming a conspicuous trace of correction.