Brainwave Assessment Model for Image Quality Evaluation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for image quality assessment using brainwaves face challenges such as difficulty in determining feature amounts for preference measurement, low precision in discriminating subjective knowledge or interests, and limited accuracy in distinguishing HDR and SDR images, due to individual differences and complex electrode-based setups.
Innovation Solution
An information processing device and method that measures brainwaves, calculates feature amounts, and constructs an assessment model to estimate subjective assessment values by associating brainwave responses with subjective feedback, using machine learning to improve precision and objectivity in image quality evaluation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If electrode-based brainwave measurement is used to assess subjective knowledge or interests, then objective measurement is achieved, but device complexity and difficulty of operation increase
Solution Approach 1:
The patent extracts and analyzes specific feature amounts from complex brainwave signals, focusing only on the relevant characteristics needed for assessment. This reduces the complexity of processing by isolating key features rather than analyzing the entire complex signal, thereby maintaining measurement precision while simplifying the analytical process.
Solution Approach 2:
The patent introduces an assessment model as an intermediary between raw brainwave measurements and subjective assessment results. This model processes the complex brainwave data and translates it into meaningful assessment outcomes, reducing the direct complexity of interpreting raw neural signals while maintaining objective measurement accuracy.
2Measurement precision
If multiple subjects are used in subjective assessment to guarantee precision, then assessment accuracy improves, but time and cost increase
Solution Approach 1:
The patent creates a computational model that copies and simulates the assessment process, allowing results to be generated without requiring multiple actual subjects. The assessment model learns from training data and can independently evaluate new inputs, replacing the need for repeated human subject assessments while maintaining precision.
Solution Approach 2:
The patent performs preliminary training of the assessment model using training datasets before actual assessment. This preliminary action prepares the model in advance, so that subsequent assessments can be conducted quickly without requiring multiple subjects for each evaluation, thereby reducing time loss while maintaining accuracy.
3Measurement precision
If brainwave feature amounts are used to determine preference, then objective assessment is achieved, but difficulty in detecting and measuring appropriate feature amounts increases
Solution Approach 1:
The patent implements a feedback mechanism where the assessment model is trained using training datasets that include both brainwave features and corresponding subjective assessments. The model continuously refines its ability to detect and measure relevant feature amounts by comparing its predictions with actual assessment results, thereby improving preference measurement accuracy while reducing the difficulty of feature detection.
Solution Approach 2:
The patent changes the parameters of the assessment model during training to optimize feature amount detection. By adjusting model parameters based on training data, the system automatically identifies the most relevant brainwave features for preference determination, reducing the manual difficulty of detecting appropriate feature amounts while maintaining high measurement precision.
Data Source
AI summary
An assessment model for enabling a subjective assessment value to be estimated from a brainwave feature amount is constructed, and assessment data which does not deviate from a subjective assessment result on the basis of an objective brainwave signal by using the assessment model can be acquired. An assessment model representing relevance between a brainwave feature amount of a subject and a subjective assessment value of the subject with respect to a stimulus is constructed by presenting the stimulus to the subject. For example, images with different image qualities and a standard image are alternately displayed on a display unit which is a stimulus presentation unit, brainwave feature amount corresponding to an image quality of a subject observing the displayed images and subjective assessment values corresponding to image qualities are acquired, and an image quality assessment model for enabling subjective assessment values to be estimated from the brainwave feature amounts is constructed by machine learning in which the brainwave feature amounts and the subjective assessment values are used as input data.