Image Scanner Edge Detection Using Reference Position Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image scanners face challenges in accurately detecting document edge positions, particularly when the scanned image contains unnecessary portions such as tabs or clips, leading to potential erroneous detections.
Innovation Solution
The image scanner employs a document table with a reference corner and an image sensor that performs line scanning while moving in a sub-scanning direction, using a document detector to repeatedly detect lateral sides by setting detection reference positions and ranges, thereby improving edge detection accuracy even with irregular sheet configurations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the image scanner detects edge positions by sequentially examining all pixels in each line from one end to the other, then it can identify pixel positions where pixel color sharply changes, but it may incorrectly detect edges when the scanned image contains unnecessary portions such as tabs or clips
Solution Approach 1:
The patent divides the edge detection process into multiple sequential stages: initial edge position estimation, determination of detection reference positions, and final edge detection within restricted ranges. This segmentation allows the system to process information in manageable steps, reducing the impact of erroneous detections from tabs or clips while maintaining overall detection accuracy.
Solution Approach 2:
The patent performs preliminary actions by first estimating initial edge positions and then determining detection reference positions based on these estimates before conducting the final edge detection. This preliminary processing establishes a framework that guides the subsequent detection process, ensuring that only relevant pixel ranges are examined and reducing the likelihood of detecting spurious edges from attachments like tabs or clips.
2Measurement precision
If the image scanner examines all pixels in the scanned image to detect edge positions, then it can identify document boundaries, but it increases the risk of erroneous detection due to influences of attached objects
Solution Approach 1:
The patent applies local quality by restricting edge detection to specific pixel ranges determined by detection reference positions rather than examining all pixels in the scanned image. This localized approach concentrates computational resources on relevant areas, improving detection precision while avoiding the complexity of processing the entire image and reducing the impact of distant objects like tabs or clips.
Solution Approach 2:
The detection process is segmented into distinct phases: initial estimation, reference position determination, and final detection within restricted ranges. Each segment has a specific function and operates on a subset of the data, which simplifies the overall process complexity compared to a monolithic approach that would need to handle all possible edge cases across the entire image.
3Ease of operation
If the image scanner uses a simple sequential pixel examination method, then the detection process is straightforward, but it cannot accurately distinguish document edges from edges of attached objects
Solution Approach 1:
The patent maintains operational simplicity by using a sequential pixel examination approach but enhances it with preliminary actions that establish detection reference positions before the main detection phase. This preliminary step provides guidance for the subsequent detection process, allowing the system to distinguish document edges from attached objects without requiring complex algorithms, thus preserving ease of operation while improving accuracy.
Solution Approach 2:
The system incorporates feedback mechanisms where detection reference positions determined from initial estimates guide the final edge detection process. This feedback loop allows the system to adjust its detection strategy based on preliminary findings, improving accuracy in distinguishing document edges from attached objects while maintaining a relatively simple overall process structure.
Data Source
AI summary
An image scanner includes a document detector configured to, each time line scanning is repeated a predetermined number of times, perform an edge detecting process to detect a position of a lateral side extending in a sub scanning direction from a non-reference corner of a sheet, in the edge detecting process performed for a first time, set a position of the non-reference corner in a main scanning direction as a detection reference position, in the edge detecting process performed for a second or later time, set a previously-detected position of the lateral side in the main scanning direction as the detection reference position, and in each edge detecting process, set a detection range in the main scanning direction on the basis of the detection reference position, and detect a specific pixel corresponding to the lateral side in the main scanning direction within the detection range in the main scanning direction.