Notebook Image Correction via Knuckle Line Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods fail to correct perspective distortions in images of two adjacent pages of a notebook or agenda captured with a digital camera, as they either leave uncorrected distortions when both pages are taken together or split hand-written notes or drawings extending over both pages when captured one at a time.
Innovation Solution
A method that processes distorted images by locating main and subordinate markers on each page, applying geometric transformations to correct perspective distortions, and selecting pixels belonging to the writing or drawing, implemented through software on devices like smartphones or tablets, using an image stretching algorithm to transform trapezoidal shapes into rectangular ones.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If two adjacent pages are taken together with a digital camera, then the capture efficiency is improved, but the perspective distortion around the knuckle line remains uncorrected
Solution Approach 1:
The patent divides the two-page correction problem into two separate single-page correction problems. By detecting the knuckle line and splitting the image into left and right page regions, each page can be independently corrected using standard perspective transformation, thereby eliminating the distortion that would otherwise remain uncorrected in the middle region.
Solution Approach 2:
The patent introduces an intermediary step of knuckle line detection and image splitting between capturing and correcting. This intermediary processing stage enables the application of simple single-page correction algorithms to achieve what would otherwise require complex two-page joint correction, resolving the contradiction between efficiency and precision.
2Manufacturing precision
If one page at a time is taken, then the perspective distortion is corrected accurately, but hand-written notes or drawings extending over two adjacent pages are not captured as a single picture
Solution Approach 1:
The patent segments the captured two-page image into left and right page regions based on knuckle line detection, then applies perspective correction to each segment independently. This allows the system to maintain accurate distortion correction while preserving drawings that span across the original two-page layout.
Solution Approach 2:
The patent creates a corrected copy of the two-page spread by processing each page separately and combining them. The result is a new image where perspective distortion is eliminated but the logical two-page structure and continuous drawings are preserved, effectively copying the intended visual layout without the distortion artifacts.
3Manufacturing precision
If a geometric transformation is applied to map boundary markers to a rectangle, then the image distortion is corrected, but the method does not work for two adjacent pages with different planes
Solution Approach 1:
The patent applies segmentation by detecting the knuckle line and separating the two-page image into independent left and right page regions. Each region is then processed individually using standard geometric transformation, making the method adaptable to two adjacent pages by breaking down the complex two-plane problem into two solvable single-plane problems.
Solution Approach 2:
The patent transitions from attempting a single three-dimensional transformation for two pages to applying two separate two-dimensional transformations to individual pages. This dimensional simplification enables the use of standard perspective correction algorithms that would otherwise be inapplicable to the more complex two-page geometry.
Data Source
AI summary
A method of processing a distorted digital image depicting two adjacent pages of a notebook is disclosed. The method can also discriminate whether the distorted picture to be processed represents either a single page or two adjacent pages of a notebook, in order to carry out either a first processing algorithm if a single page has been captured or the method of processing a distorted digital image depicting two adjacent pages. The methods may be implemented via software with a PC, a smartphone or a tablet. A hardback notebook and a hardback agenda are also disclosed.