AI Image Stabilization Using Focal-Point Frame Cropping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Video recordings are often marred by blurriness due to camera movement during recording, leading to reduced production quality and potential motion sickness, with existing mechanical and electronic image stabilization techniques being inadequate.
Innovation Solution
Intelligent digital image stabilization methods that utilize AI and ML to detect focal points, determine vector differences between frames, and crop images to maintain a fixed frame viewpoint, thereby stabilizing the video.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If mechanical image stabilization techniques using gyroscopes and mechanical arms are used, then camera movement compensation is improved, but device complexity and cost increase
Solution Approach 1:
The patent replaces mechanical stabilization systems (gyroscopes, mechanical arms) with an electronic/image processing-based stabilization system. The method uses sequential image processing, vector difference calculation, and digital cropping to achieve stabilization without mechanical components, thereby reducing device complexity while maintaining or improving stabilization effectiveness.
Solution Approach 2:
The patent creates a virtual copy of the stabilization effect through digital image processing. Instead of physically moving the camera or sensor to compensate for movement, the system processes the captured images to simulate a stabilized viewpoint, achieving the same effect as mechanical stabilization but through computational methods.
2Measurement precision
If electronic image stabilization relying on gyroscopes is used, then movement measurement is improved, but device complexity increases
Solution Approach 1:
The patent substitutes physical measurement devices (gyroscopes) with an image-based measurement approach. By detecting focal points in sequential images and calculating vector differences, the system determines camera movement purely through image processing, eliminating the need for mechanical sensors while achieving comparable or superior measurement precision.
Solution Approach 2:
The system uses the captured images themselves to measure camera movement, rather than requiring separate measurement devices. The images serve dual purposes: as both the subject to be stabilized and as the reference data for determining the stabilization parameters, reducing hardware requirements.
3Reliability
If focal points are detected and images are cropped to maintain fixed frame viewpoint, then image stabilization is improved, but image area is reduced
Solution Approach 1:
The patent applies partial action by cropping only the necessary portions of the image that contain the stabilized focal point, rather than stabilizing the entire image frame. This approach maintains the stability of the subject of interest while minimizing the loss of overall image area, achieving a balance between stabilization effectiveness and image utilization.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Example implementations include a method, apparatus and computer-readable medium of intelligent digital image stabilization, comprising obtaining a first full image from a sensor. The implementations further include detecting at least one focal point of the first full image. Additionally, the implementations further include obtaining a second full image from the sensor including the at least one focal point. Additionally, the implementations further include determining a vector difference between the at least one focal point within the second full image and the at least one focal point within the first full image. Additionally, the implementations further include cropping the second full image based on the vector difference to generate a display image. Additionally, the implementations further include displaying the display image.