Image Interpolation Using High Cutoff Low-Pass Filter for Periodic Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image interpolation devices are insufficiently accurate, particularly when interpolating periodic data scanned from documents, as they struggle to accurately fill in missing pixels at the boundaries between sensor ICs in contact image sensors.
Innovation Solution
An image interpolation device employing an improved low-pass filter with a high cutoff frequency is used to effectively interpolate pixels in periodic data, ensuring accurate interpolation of frequencies lower than the cutoff frequency by utilizing a specific low-pass interpolation filter and mean-preserving interpolators.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional pixel interpolation methods (mean, regression line, quartic curve) are used, then the interpolation can be calculated simply, but the accuracy is insufficient particularly for periodic data
Solution Approach 1:
The patent changes the parameters of the low-pass filter by setting a high cutoff frequency (Fc) that is higher than the maximum frequency component of the periodic data. This parameter change allows the filter to preserve high-frequency periodic components while still filtering out noise, thereby improving interpolation accuracy for periodic data without requiring complex adaptive methods
Solution Approach 2:
The patent segments the interpolation process into two distinct parts: (1) a low-pass filter interpolation that handles the general smoothing and noise reduction, and (2) a mean-preserving interpolation that specifically targets periodic data regions. This segmentation allows each method to specialize in handling different aspects of the interpolation problem, improving overall accuracy while keeping individual method complexities manageable
2Measurement precision
If a low-pass filter with high cutoff frequency is used, then periodic data can be interpolated properly, but high-frequency noise may be preserved
Solution Approach 1:
The patent applies different interpolation qualities to different regions of the data. The mean-preserving interpolation is selectively applied to regions identified as containing periodic data, while the low-pass filter handles other regions. This local differentiation allows high-frequency periodic components to be preserved where needed while still providing noise filtering where appropriate
Solution Approach 2:
The patent uses a marking circuit that analyzes the interpolated data and generates marking data to identify periodic regions. This feedback mechanism allows the system to detect where periodic patterns exist and apply the appropriate interpolation method accordingly, ensuring that high-frequency noise is not preserved in non-periodic regions while periodic components are maintained in periodic regions
Data Source
Figure 1~2
Figure 3(a)~3(b)
Figure 4(a)~5
AI summary
An image interpolation device comprises a data storage unit (2) for storing the data in the neighborhood of a missing pixel, a filter interpolation unit (3) for performing a multiply-add computation on the data output from the data storage unit, a limiter processing unit (4) for limiting the data output from the filter interpolation unit to within the range of the data output from the data storage unit, and an interpolation data insertion unit (5) for interpolating the data output from the limiter processing unit into a position corresponding to the missing pixel in the data output from the data storage unit. The image interpolation device is capable of increasing the interpolation accuracy of missing pixels, especially in the interpolation of scanned document data including periodic data.