Virtual Pixel Brightness Determination via Homogeneous Noise Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing methods, such as interpolation, fail to generate image data with characteristics corresponding to an image sensor with a changed pixel size and number, leading to undesirable changes in image quality and noise distribution, particularly when changing pixel size non-integers and affecting applications like digital cameras.
Innovation Solution
An image processing apparatus that determines brightness values for virtual second pixels using local filters with specific filter coefficients, ensuring the sum of squared coefficients is constant across filters, allowing for spatially homogeneous noise transmission and adherence to the EMVA standard 1288 pixel model, thereby achieving physically sensible pixel size changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If existing interpolation methods are used to generate image data with changed pixel size, then the pixel size can be changed, but the image quality and noise distribution become distorted and do not correspond to real pixel characteristics
Solution Approach 1:
The patent applies parameter changes by modifying the filter coefficients based on the relative pixel size. The filter coefficients are adjusted according to the ratio between original and target pixel sizes, ensuring that the interpolation process produces results that match the characteristics of real pixels with the desired size. This resolves the contradiction by dynamically adapting the processing parameters to maintain image quality accuracy while enabling pixel size changes.
Solution Approach 2:
The patent performs preliminary determination of filter coefficients before the actual interpolation process. By pre-calculating the appropriate filter coefficients based on the desired pixel size transformation, the system ensures that the interpolation will produce accurate results. This preliminary action prevents image quality distortion and maintains noise distribution fidelity throughout the pixel size change process.
2Device complexity
If standard interpolation is applied, then computational simplicity is maintained, but noise transmission becomes non-homogeneous and quantum efficiency characteristics are not preserved
Solution Approach 1:
The patent applies local quality by using local filters that are applied at each pixel position independently. Each local filter is tailored to the specific neighborhood of pixels being interpolated, allowing the noise transmission to be homogeneous across the entire image while maintaining computational efficiency. This resolves the contradiction by making the processing locally adaptive rather than globally complex.
3Adaptability or versatility
If pixel size is changed using conventional methods, then different camera variants can be simulated, but the quantum efficiency and pixel model adherence are compromised
Solution Approach 1:
The patent achieves universality by creating a single image processing apparatus that can simulate multiple camera variants with different pixel sizes. By using the same apparatus with dynamically adjusted filter coefficients, the system can produce image data that adheres to the EMVA 1288 pixel model for any desired pixel size, maintaining quantum efficiency accuracy across all simulated variants without requiring separate hardware for each configuration.
Data Source
AI summary
The invention relates to an image processing apparatus for processing image data of an image sensor with a regular arrangement of first pixels, wherein the image processing apparatus is configured to determine a brightness value for each of two or more virtual second pixels of the same size at different intermediate positions between the first pixels, wherein the determination of the respective brightness value comprises an interpolation of the pixels of a neighborhood of the respective intermediate position, by means of a local filter, wherein each of the local filters comprises a plurality of filter coefficients, wherein for at least one of the local filters more than one of the filter coefficients is unequal to zero, and wherein the sum of the squared filter coefficients for each of the local filters is equal to a constant value, which according to a first condition is the same for all local filters.