Customized Image Reprocessing System Using Machine Learning

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Users often capture unflattering selfies in low light conditions due to the lack of flash on front-facing cameras, leading to images that are not shared or discarded, as existing image editing processes are cumbersome and unsatisfying, especially when applying corrective filters that do not account for individual preferences and varying light conditions.

Innovation Solution

A customized image reprocessing system powered by machine learning, which automatically reprocesses images on a pixel-by-pixel level using sensor data and user-preferred flash calibration parameters, adjusting parameters such as brightness and color temperature to enhance image quality.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If traditional image editing processes are used, then users can manually adjust images, but the process is cumbersome and unsatisfying

Engineering Contradiction:

Improveimage editing processVSAvoidimage processing efficiency

Core Design Contradiction:

Ease of operationVSProductivity

Solution Approach 1:

The patent replaces manual mechanical image editing operations with an automated machine learning-based reprocessing system. The system automatically analyzes captured images and applies appropriate processing parameters without requiring users to manually adjust settings, thereby eliminating the cumbersome nature of traditional editing while maintaining high processing efficiency

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system performs self-service by automatically determining the best processing parameters for each image based on its characteristics. The machine learning model independently analyzes the image content, lighting conditions, and subject matter to select optimal processing parameters, eliminating the need for user intervention in the editing process

Inventive Principle:

Principle #25Self-service

2Reliability

If corrective filters are applied to images, then image quality can be improved, but the filters do not account for individual preferences and varying light conditions

Engineering Contradiction:

Improveimage quality enhancementVSAvoidadaptation to individual preferences

Core Design Contradiction:

ReliabilityVSAdaptability or versatility

Solution Approach 1:

The patent applies local quality by tailoring image processing parameters to specific local characteristics of each image. The machine learning model analyzes individual image properties such as lighting conditions, subject type, and composition to determine optimal processing parameters, ensuring that each image receives customized treatment rather than applying a generic filter to all images

Inventive Principle:

Principle #3Local quality

Solution Approach 2:

The system dynamically changes processing parameters based on image characteristics and user preferences. The machine learning model adjusts parameters such as brightness, contrast, color temperature, and sharpness according to the specific conditions of each captured image, enabling adaptive enhancement that accounts for varying light conditions and individual user preferences

Inventive Principle:

Principle #35Parameter changes

3Device complexity

If front-facing cameras are used without flash, then device portability is maintained, but image quality in low light conditions deteriorates

Engineering Contradiction:

Improvecamera system designVSAvoidimage lighting quality

Core Design Contradiction:

Device complexityVSIllumination intensity

Solution Approach 1:

The patent substitutes the mechanical flash illumination system with a computational imaging approach. Instead of adding physical lighting hardware to the front-facing camera, the system uses machine learning-based image reprocessing to compensate for poor lighting conditions, maintaining device simplicity while improving image quality in low light environments

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The machine learning model acts as an intermediary between the captured image and the final processed output. It analyzes the relationship between the original image characteristics and desired output quality, then applies appropriate processing parameters to bridge the gap between poor lighting conditions and acceptable image quality without requiring additional hardware

Inventive Principle:

Principle #24Intermediary (Mediator)

Data Source

PatentUS11770621B2Customized image reprocessing system using a machine learning model

Publication Date: 2023.09.26 SNAP INC

AI summary

The technical problem of automatically reprocessing an image captured by a camera in a manner that produces a personalized result is addressed by providing a customized image reprocessing system powered by machine learning techniques. The customized image reprocessing system is configured to automatically reprocess an image on a pixel level using a machine learning model that takes, as input, the image represented by pixel values, sensor data detected by the digital sensor of a camera at the time the image was captured, and, also, flash calibration parameters previously generated for that specific user.