Image Processor Noise Smoothing via Dynamic Random Pixel Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing techniques struggle to reduce noise while maintaining the frequency distribution characteristics of noise in images, often resulting in uneven noise distribution and degradation of image patterns.
Innovation Solution
An image processor that calculates a multiplication coefficient based on the similarity between regions in an image, using random numbers to perform noise smoothing, ensuring that both low-frequency and high-frequency noise components are uniformly preserved, thereby maintaining the natural appearance of images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If noise smoothing is performed using conventional techniques (spatial filtering or bilateral filtering), then noise is reduced, but the frequency distribution characteristics of the noise are altered and image patterns may be degraded
Solution Approach 1:
The patent applies dynamics by making the selection of reference pixels dynamic and random rather than fixed or systematic. Specifically, reference pixels are selected randomly from the surrounding pixels of the target pixel, and this random selection is performed differently for each target pixel. This dynamic random selection preserves the natural frequency distribution of noise while still achieving noise reduction through the weighted averaging process.
Solution Approach 2:
The patent changes the parameter of reference pixel selection from fixed spatial arrangements or intensity-based selections to random selections. By changing how reference pixels are chosen (using random numbers to determine which surrounding pixels to use), the patent maintains the statistical properties of noise while reducing its visual impact. The multiplication coefficient also dynamically adjusts based on intensity differences between target and reference pixels.
2Ease of operation
If reference pixels are selected systematically (e.g., along axes or based on intensity similarity), then processing is more structured, but noise distribution becomes uneven and frequency characteristics are lost
Solution Approach 1:
Instead of systematically selecting reference pixels based on spatial arrangement or intensity similarity (conventional approach), the patent inverts the approach by randomly selecting reference pixels. This inversion from systematic to random selection prevents the creation of artificial patterns in the noise distribution while still achieving effective noise reduction through the weighted averaging process.
3Object-affected harmful factors
If strong noise smoothing is applied to reduce noise effectively, then noise level decreases, but image patterns and natural appearance are degraded
Solution Approach 1:
The patent changes the parameter of reference pixel selection to random selection, which fundamentally alters how noise smoothing is applied. This random approach prevents the creation of artificial patterns that would result from systematic filtering, thereby preserving image patterns and natural appearance even while achieving effective noise reduction through the weighted averaging process.
Data Source
AI summary
An image processor includes: a first calculator configured to calculate similarity between a first region including a target pixel and a second region including a reference pixel and calculate a multiplication coefficient which increases as the similarity increases; a second calculator configured to calculate a random number; a sum-of-products arithmetic unit configured to multiply a pixel value of each reference pixel by the multiplication coefficient and the random number and compute a sum of the products; a coefficient summation unit configured to multiply the multiplication coefficient by the random number and compute a sum of the products; and a division unit configured to divide the result of the sum-of-products arithmetic unit by the result of the coefficient summation unit.