Adaptive Bit-Depth Adjustment Circuitry for Display Banding Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Electronic displays face challenges in maintaining perceived image quality due to banding visual artifacts caused by fixed bit-depth image data, which limits the number of grayscale values, especially in gradient content, leading to reduced image quality and increased implementation costs associated with larger storage capacities for bit-depth adjustment.
Innovation Solution
Implementing a bit-depth adjustment circuitry using machine learning techniques, such as convolutional neural networks, to adaptively adjust the bit-depth of image data in smaller pixel windows, reducing the need for large storage buffers and enabling selective bit-depth adjustments based on perceivability thresholds, thereby improving operational efficiency and image quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional gradient calculation and banding removal techniques are used, then banding artifacts can be removed, but large image pixel windows are required which increase input buffer storage capacity and implementation cost
Solution Approach 1:
The patent replaces traditional mechanical gradient calculation methods with a machine learning-based neural network approach. The neural network is trained to predict banding artifacts directly from image data, eliminating the need for explicit gradient calculations and large analysis windows, thereby reducing input buffer storage requirements while maintaining banding removal effectiveness
Solution Approach 2:
The patent changes the fundamental parameter of analysis window size by using a trained neural network that can accurately detect banding artifacts in smaller windows. The neural network's learned features enable effective banding detection with reduced spatial context, allowing implementation with smaller input buffers compared to traditional gradient-based methods that require larger windows for accurate gradient calculation
2Reliability
If bit-depth of image data is increased to reduce banding artifacts, then image quality improves, but storage capacity requirements and implementation cost increase
Solution Approach 1:
The patent applies local quality by using the neural network to selectively identify and correct banding artifacts only in specific regions of the image where they are present. Instead of uniformly increasing bit-depth across the entire image, the system processes only affected areas, reducing overall storage capacity requirements while maintaining image quality in critical regions
Solution Approach 2:
The patent implements partial action by applying bit-depth adjustment only to image regions where banding artifacts are detected by the neural network, rather than processing the entire image. This selective approach reduces the total amount of data that requires increased storage capacity while still effectively removing visible banding artifacts where they occur
3Measurement precision
If larger image pixel windows are used for bit-depth adjustment, then banding detection accuracy improves, but processing time and computational complexity increase
Solution Approach 1:
The patent applies preliminary action by pre-training the neural network offline to learn banding artifact patterns and detection features. This pre-training phase performs the computationally intensive work of analyzing large datasets and learning optimal detection parameters, so that during actual runtime, the trained network can quickly and accurately detect banding in smaller windows without requiring extensive real-time computation
Solution Approach 2:
The patent substitutes traditional mechanical gradient calculation and analysis methods with a trained neural network that has already learned optimal detection patterns. This substitution enables accurate banding detection in smaller windows during runtime, reducing processing time compared to traditional methods that would require analyzing larger windows with explicit gradient calculations
Data Source
AI summary
Systems and methods for improving perceived image quality with reduced implementation associated cost and/or improved operational efficiency. A display pipeline includes an input buffer that stores input image data corresponding with an image pixel window, in which the input image data has a first bit-depth and includes image data corresponding with an image pixel in the image pixel window. The display pipeline includes bit-depth adjustment circuitry, which includes a neural network that operates based on a set of bit-depth adjustment parameters to process the input image data to determine whether banding greater than a perceivability threshold is expected to result when the image is displayed directly using the input image data with the first bit-depth and to process the image data corresponding with the image pixel to expand the image data from the first bit-depth to a second bit-depth when the banding visual artifact is greater than the perceivability threshold.