Noise Determination Using Video Type and Pixel Correlation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for removing noise from digital broadcasts often incorrectly distinguish between texture areas and noise, particularly in cases where low-level texture is similar to noise or where there is a significant level difference with neighboring pixels.
Innovation Solution
A noise determining apparatus and method that analyze video characteristics such as resolution, aspect ratio, and frame rate to determine the type of video, and use correlation calculations between pixel and search windows to accurately assess noise levels and presence, employing a pixel window and a search window to differentiate noise from texture.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If noise removal filtering is applied to digital broadcast images, then noise signals are reduced, but texture areas of low level are incorrectly determined to be noise and edge areas are damaged
Solution Approach 1:
The patent applies dynamics by making the noise determination process adaptive to different video types. The system dynamically adjusts noise determination thresholds and parameters based on whether the video is HD or SD, and whether resolution conversion has been applied. This dynamic adaptation allows the system to maintain high texture discrimination accuracy across different video conditions while effectively removing noise.
Solution Approach 2:
The patent changes parameters by introducing video type classification (HD/SD, resolution converted/not converted) as additional determination criteria. By changing the determination parameters from simple pixel value comparison to a multi-parameter assessment including video type, correlation values, and noise levels, the system achieves more accurate texture area discrimination while maintaining noise removal effectiveness.
2Device complexity
If pixel value-based texture area determination is used, then processing is simple, but incorrect discrimination occurs between low-level texture and noise
Solution Approach 1:
The patent segments the noise determination process into distinct stages: video type analysis, correlation calculation, noise level determination, and final noise presence determination. By segmenting the process and adding the video type classification step, the system maintains reasonable complexity while dramatically improving discrimination accuracy between texture and noise.
Solution Approach 2:
The patent introduces video type classification as an intermediary step between simple pixel analysis and final noise determination. This intermediary assessment of video characteristics (HD/SD, resolution conversion status) provides additional context that mediates the determination process, enabling more accurate texture-noise discrimination without excessive complexity.
3Productivity
If noise determination is based on pixel values alone, then processing is fast, but noise with great level difference from neighboring pixels is incorrectly identified as texture
Solution Approach 1:
The patent applies preliminary action by performing video type analysis and correlation calculation before final noise determination. By pre-assessing video characteristics and calculating correlations between pixels in advance, the system prepares determination criteria that speed up the final noise detection while improving accuracy for pixels with large level differences from neighbors.
Data Source
AI summary
A noise determining apparatus is provided. The noise determining apparatus includes a video determiner which determines type of video according to a pre-set criterion, a noise level determiner which determines a level of noise with reference to output from the video determiner, and a noise determiner which determines presence or absence of noise with reference to output from the noise level determiner. Accordingly, incorrect discrimination between a texture area of low level which is similar to noise and noise having a great level difference with respect to neighboring pixels is reduced.