Panoramic Video Occlusion via Component Image Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video monitoring systems lack effective methods to protect personal privacy during image capturing or processing, as they do not adequately address the need to obscure specific regions in panoramic images formed from multiple component images.
Innovation Solution
A system comprising multiple imaging devices, a storage device, and a processor that captures component images, combines them into panoramic images, and determines occlusion regions to block privacy areas by adjusting parameters based on combination coefficients and occlusion region information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If multiple component images are combined to form a panoramic image for comprehensive monitoring, then the monitoring coverage is improved, but the personal privacy security deteriorates due to inability to effectively obscure specific regions
Solution Approach 1:
The panoramic image is segmented into multiple component images, and occlusion processing is applied independently to each component image based on its specific occlusion regions. This allows privacy protection to be targeted at specific regions while maintaining monitoring coverage in other areas.
Solution Approach 2:
Different occlusion processing strategies are applied to different component images based on their local characteristics. The occlusion region in each component image is determined by transforming coordinates from the panoramic image space, allowing localized privacy protection without affecting the entire panoramic view.
2Measurement precision
If occlusion regions are determined in the panoramic image and transformed to component images, then the privacy protection precision is improved, but the system complexity increases due to coordinate transformation and parameter calculation
Solution Approach 1:
The combination coefficients (cropping width, cropping height, maximum height distortion) are pre-calculated and stored for each component image. During occlusion processing, these pre-computed parameters are directly used in coordinate transformation, avoiding repeated complex calculations and reducing real-time processing complexity.
Solution Approach 2:
The panoramic image serves as an intermediary coordinate system that simplifies occlusion region definition. Occlusion regions are first determined in the panoramic image space using unified coordinates, then transformed to individual component image spaces using pre-computed combination coefficients, making the overall process more manageable.
3Manufacturing precision
If combination coefficients including cropping width, cropping height, and maximum height distortion are used to transform occlusion regions, then the occlusion effect is improved, but the calculation complexity increases
Solution Approach 1:
All combination coefficients (cropping width, cropping height, maximum height distortion) are pre-calculated during the panoramic image formation stage and stored for later use. When processing occlusion regions, these pre-computed parameters are directly applied without recalculation, significantly reducing the computational burden during actual privacy protection operations.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present disclosure relates to systems and methods for generating an occlusion region in an image. The system includes a plurality of imaging devices, a storage device and at least one processor in communication with the storage device, wherein each of the plurality of imaging devices is configured to capture a component image presenting a scene. When executing the instructions, the at least one processor is configured to cause the system to determine a first occlusion region in an image formed based on at least one component image. The at least one processor is configured to determine how to generate a second occlusion region in at least one of the plurality of component images based on the area of the first occlusion region and an area threshold.