Image Signal Processor Parameter Tuning Using Machine Learning

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing methods for tuning image signal processors rely heavily on subjective human evaluation, making it difficult to objectively and precisely adjust parameters for optimal image quality.

Innovation Solution

A machine learning model is trained to predict optimal parameter values for an image signal processor by capturing sample images, generating sample scores, and adjusting weights based on evaluation items, thereby reducing subjective intervention and improving image quality objectively.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If human evaluation is used to tune image signal processors, then flexibility in judgment is maintained, but objectivity and precision are compromised

Engineering Contradiction:
Improveparameter tuning precisionVSAvoidautomation level
Core Design Contradiction:
Measurement precisionVSExtent of automation

Solution Approach 1:

The patent replaces human evaluation (mechanical/subjective system) with a machine learning-based automated evaluation system. The ML model processes image data and outputs numerical scores for multiple evaluation items, eliminating subjective human judgment while maintaining comprehensive evaluation capabilities. This substitution achieves both high precision and full automation.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The machine learning model serves as an intermediary between the image signal processor and the evaluation system. Instead of direct human evaluation, the ML model acts as a mediator that objectively assesses image quality based on multiple parameters, transforming subjective human judgment into objective numerical scores that can drive automated parameter optimization.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Manufacturing precision

If multiple parameters are adjusted to optimize image quality, then image generation quality improves, but system complexity increases

Engineering Contradiction:
Improveimage quality optimizationVSAvoidparameter adjustment complexity
Core Design Contradiction:
Manufacturing precisionVSDevice complexity

Solution Approach 1:

The patent systematically changes multiple parameters (exposure, white balance, sharpness, noise reduction, etc.) through automated optimization. The machine learning model evaluates images generated with different parameter combinations and provides numerical scores, enabling the system to automatically adjust parameters to optimal values without manual intervention. This achieves high image quality while managing complexity through algorithmic automation.

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The system implements a feedback loop where images generated with specific parameters are evaluated by the machine learning model, and the numerical scores feed back into the optimization process. This feedback mechanism enables continuous automatic adjustment of parameters, simplifying the complexity management by providing an automated closed-loop control system that converges on optimal parameter values.

Inventive Principle:
Principle #23Feedback

Data Source

PatentEP3742389B1Predicting optimal values for parameters used in an operation of an image signal processor using machine learning
Publication Date: 2026.04.22 SAMSUNG ELECTRONICS CO LTD
  • EP3742389B1 patent drawingFigure 1
  • EP3742389B1 patent drawingFigure 2
  • EP3742389B1 patent drawingFigure 3

AI summary

Disclosed is a method of predicting optimal values for a plurality of parameters used in an operation of an image signal processor, the method including: inputting initial values for the plurality of parameters to a machine learning model having an input layer, corresponding to the plurality of parameters, and an output layer corresponding to a plurality of evaluation items extracted from a resultant image generated by the image signal processor; obtaining evaluation scores for the plurality of evaluation items using an output of the machine learning model; adjusting weights, applied to the plurality of parameters, based on the evaluation scores; and determining the optimal values using the adjusted weights.