Wide Angle Camera Viewing Frustum Selection Using Hierarchy of Interest
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current wide angle surveillance cameras with fisheye or anamorphic lenses suffer from image distortion, and existing systems for determining a viewing frustum are ineffective in consistently selecting the most interesting area for the viewer, leading to issues like 'jitter' and improper selection.
Innovation Solution
A camera system employing multiple detection sub-systems and a Hierarchy of Interest (Hol) to select and de-warp the viewing frustum, ensuring that the most interesting parts of the image are transmitted to the viewer, with the ability to adapt based on feedback.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If a wide angle lens is used to capture a large viewing area, then the coverage area is improved, but image distortion increases
Solution Approach 1:
The patent divides the captured image into multiple regions of interest (ROIs) based on detected objects or events. Each ROI is processed independently through de-warping operations, allowing selective correction of distortion in areas containing important content while preserving the wide-angle perspective in other areas.
Solution Approach 2:
The patent applies de-warping operations selectively to specific regions of the image rather than uniformly across the entire image. The degree and type of de-warping is adjusted locally based on the content detected in each region, optimizing the balance between distortion correction and perspective preservation for each area.
2Productivity
If motion detection-based algorithms are used to select viewing frustum, then the system responds to active areas, but it causes jitter when motion occurs in multiple areas
Solution Approach 1:
The patent implements a feedback mechanism where the system continuously monitors motion in multiple regions and adjusts the viewing frustum selection based on the aggregate feedback from all detection algorithms. When motion is detected in multiple areas, the system evaluates the significance and hierarchy of each motion event to determine the appropriate frustum, preventing rapid switching and jitter.
Solution Approach 2:
The patent changes the parameters used for frustum selection by incorporating multiple detection algorithms (person detection, object detection, motion detection) with different sensitivity thresholds and priority weights. This multi-parameter approach allows the system to distinguish between significant and insignificant motion, stabilizing the viewing frustum selection even when multiple motion events occur.
3Measurement precision
If person detection-based algorithm is used to select viewing frustum, then the system focuses on areas with people, but it fails when no people are present
Solution Approach 1:
The patent implements a universal detection framework that incorporates multiple detection algorithms (person detection, object detection, motion detection) that can all function independently. When person detection fails to find targets, the system seamlessly transitions to using object detection or motion detection algorithms, ensuring continuous and accurate viewing frustum selection across diverse scenarios without relying on a single detection method.
Data Source
AI summary
A novel system and method for determining a viewing frustum for a wide angle camera system employs multiple detection sub-systems to detect the presence or occurrence of events and/or objects of interest within the entire captured field of view of the system and processes the results of those detection sub-systems in accordance with a Hierarchy of Interest to select a viewing frustum for providing a de-warped view of interest to a viewer. The Hierarchy of Interest can be predefined, or selected from a set of Hierarchies of Interest and/or can be adaptive is response to feedback from a viewer.