Panoramic Image Spreading via Human Feature Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional panoramic cameras, whether with fisheye lenses or not, suffer from image distortion and require complex cutting, intercepting, and stitching processes, often resulting in human faces being cut and split during the formation of two-dimensional panoramic images, limiting their application and industrial value.
Innovation Solution
A video conference panoramic image spreading method that uses bodily human features recognition to identify and avoid dividing points, allowing for the selection of a starting point for cutting and stitching that does not split human features, enabling the formation of a two-dimensional panoramic image without the need for reference objects or additional instruments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If cutting and stitching operations are performed to form a two-dimensional panoramic picture, then the panoramic image can be displayed in a standard format, but human faces and bodies may be cut and split across opposite sides
Solution Approach 1:
The patent performs preliminary detection of human faces and bodies in the panoramic image before executing cutting and stitching operations. By identifying the positions and extents of human features in advance, the system can select cutting lines that avoid dividing these features, thereby preventing face splitting while still achieving the desired two-dimensional panoramic picture format
Solution Approach 2:
The patent applies different processing strategies to different regions of the panoramic image. Specifically, regions containing human faces or bodies are identified and given special consideration during the cutting process, while other regions can be processed normally. This localized approach ensures that human features maintain their integrity while the rest of the image is properly formatted
2Ease of operation
If a reference object is set up at a selected angular position to provide a reference point for cutting and stitching, then the cutting starting point can be determined, but the installation and application range of the panoramic camera is significantly limited
Solution Approach 1:
The patent enables the panoramic camera system to automatically determine cutting starting points by detecting human faces and bodies within the captured image. This self-service capability eliminates the need for external reference objects or manual configuration, allowing the system to adapt to any camera installation position and orientation while still performing accurate cutting and stitching operations
Solution Approach 2:
The patent replaces the mechanical approach of using physical reference objects with an image processing-based solution. By using computational algorithms to detect human features and determine cutting lines, the system achieves the same functional result without requiring any physical reference markers, thereby greatly increasing installation flexibility
3Manufacturing precision
If seats for photographed subjects are particularly arranged to avoid coverage in front of the reference object, then human faces can avoid being cut and split, but the locations and quantity of seats are subjected to unnecessary constraints
Solution Approach 1:
The patent performs preliminary detection of human faces and bodies in the panoramic image before executing cutting operations. By identifying the positions of all photographed subjects in advance, the system can automatically determine cutting lines that avoid dividing any human features, thereby maintaining face integrity while requiring no special seat arrangements
Solution Approach 2:
Instead of arranging seats to avoid certain areas (traditional approach), the patent inverts the problem by keeping seats anywhere and instead adjusting the cutting line selection to avoid human features. This inversion of the solution approach maintains full flexibility in seat arrangement while still achieving the goal of preventing face splitting
Data Source
AI summary
A video conference panoramic image spreading method is provided, in which an image capturing (acquiring) step is applied by using at least one panoramic camera to capture at least one video conference panoramic picture or image and a bodily human features recognition step is applied to achieve bodily human features recognition for recognizing each of conferee photographed subjects in the video conference panoramic picture or image, and then, based on the result of bodily human features recognition, steps of searching out a dividing point for a two-dimensional (2D) spread picture, image cutting and two-dimensional picture spreading, and stitching are carried out so as to avoid, in a real-time manner, conflicts of cutting of a human body in the conference panoramic picture or image based on the dividing point for two-dimensional spreading and correct the dividing point, for acquiring a two-dimensional panoramic picture that instantaneously reflects and completely and correctly corresponds to a primitive video conference panoramic picture.