Image Stabilization Control Using Machine Learning Shake Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image stabilizing devices in electronic devices, such as cameras and smartphones, face challenges in accurately determining the image capturing state, leading to inefficiencies in image stabilization, particularly due to delays in processing camera shake information after a capturing instruction is given.
Innovation Solution
The implementation of a machine learning model that acquires and processes shake information before a capturing instruction is given, allowing for pre-emptive image stabilization control, utilizing both angular velocity and acceleration sensors to determine the type of shake and communicate this information for effective stabilization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If image stabilization control is performed after capturing instruction using conventional methods, then image stabilization can be applied, but determination accuracy of image capturing state is insufficient and delays occur
Solution Approach 1:
The system performs preliminary determination of the image capturing state by analyzing shake information from angular velocity and acceleration sensors before the capturing instruction is given. This advance determination allows the image stabilization control to be prepared in advance, eliminating processing delays and improving both determination accuracy and response time.
2Measurement precision
If machine learning model is used to process shake information, then determination accuracy improves, but processing time increases
Solution Approach 1:
The machine learning model processes shake information in advance before the capturing instruction is given, performing the computationally intensive determination work during the preview period. This preliminary processing ensures that when capturing is instructed, the determination is already complete, thus improving accuracy without adding delay to the actual capturing process.
Solution Approach 2:
The system skips the machine learning processing step during actual capturing by having already completed the determination beforehand. The processed results are stored and directly used for image stabilization control, allowing the system to rush through the capturing process without repeating the time-consuming machine learning analysis.
3Measurement precision
If shake information is processed after capturing instruction, then real-time stabilization is achieved, but accuracy is compromised due to insufficient data
Solution Approach 1:
The system collects and processes shake information during the preview period before capturing is instructed, accumulating sufficient data for accurate determination. This preliminary action ensures both high accuracy through adequate data collection and fast response speed since the analysis is completed before capturing begins.
Data Source
AI summary
Ac device includes an acquisition unit configured to acquire first information regarding shake, a calculation unit configured to input the first information to a machine learning model and output second information regarding a type of the shake, and a first control unit configured to control an image stabilization using the second information. By using the second information output from the calculation unit based on the first information before a capturing instruction is given, the first control unit controls an image stabilization after the capturing instruction is given.