Video Frame Cropping for Stationary Camera Target Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional video collection methods using pan-tilt-zoom mechanisms for follow shots are costly.
Innovation Solution
A video collection method for electronic devices that does not require pan-tilt-zoom, allowing the camera to track and maintain a target object in a specified region by cropping and adjusting the video frame based on the target's location, using human body detection and user operations to determine and follow the object.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If pan-tilt-zoom mechanisms are used to implement follow shot, then the camera can rotate and track the target object, but the cost increases significantly
Solution Approach 1:
The patent replaces the mechanical pan-tilt-zoom system with a digital image processing system. The camera remains stationary while software algorithms detect target objects in each frame and dynamically adjust the cropping region to follow the target's movement, achieving the same tracking effect without mechanical rotation components
Solution Approach 2:
Instead of physically moving the camera to track the target, the patent creates a digital copy or representation of the target's position and movement trajectory through image recognition and processing. The system captures the target's location in each frame and uses this information to generate a moving cropping window that follows the target, effectively copying the tracking behavior through data rather than physical motion
2Reliability
If the camera shooting angle is rotated to follow the target, then the target remains centered in the frame, but the device complexity increases
Solution Approach 1:
The patent substitutes the mechanical camera rotation system with a digital image cropping and processing system. The camera lens remains fixed while the electronic image sensor data is processed to extract and reframe the region containing the target object, achieving centering accuracy through software rather than mechanical adjustment
Solution Approach 2:
The patent transitions from physical spatial rotation (changing the camera's orientation in 3D space) to digital image space transformation (changing the displayed image frame). By operating in the digital image dimension rather than the physical mechanical dimension, the system achieves target centering without increasing device structural complexity
Data Source
Figure 1A
Figure 1B
Figure 2
AI summary
The present invention relates to the field of image processing, and discloses a video processing method, an electronic device, a computer-readable storage medium, a computer program product, and a chip, to resolve a technical problem in the conventional technology that a rotating camera needs to be set to track a user in a video call. The method includes: obtaining a first video frame, determining that the first video frame includes at least one target object, determining a cropping box by using the at least one target object as a center, and obtaining content in the cropping box, and displaying the content as a second video frame. The method may be applied to an artificial intelligence device, and the method is related to a technology such as deep learning.