Ontology-Based Saliency for Real-Time Change Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Visual-based autonomous change detection using unmanned vehicles is challenging due to variations in illumination, lighting, and environmental conditions, and existing systems lack the ability to efficiently detect changes in real-time using ontology-based saliency.
Innovation Solution
A system and method that utilize ontology-based saliency to detect real-time changes in objects within an unknown environment. This involves capturing images, identifying salient regions based on predefined ontology, calculating saliency, registering consecutive images using Gaussian pyramids, and comparing saliency maps to detect changes using probability differences.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional saliency detection using image attributes (color, orientation, texture, motion) is used, then visual change detection can be performed, but computation time increases and real-time detection becomes difficult due to complex scenarios requiring weight re-adjustment of multiple attributes
Solution Approach 1:
The patent extracts and utilizes only the motion attribute from traditional saliency detection, discarding other attributes like color, orientation, and texture. This extraction of the essential motion component enables real-time change detection while maintaining accuracy, as motion is the primary indicator of change in the monitored environment.
Solution Approach 2:
The patent segments the video feed into individual frames and processes them sequentially, comparing each frame with the previous one to detect motion. This frame-by-frame segmentation allows for efficient computation of motion vectors and saliency maps, reducing overall computation time while maintaining detection precision.
2Productivity
If fixed criterion for narrowing down regions of interest is used, then fast computation during navigation is achieved, but adaptability to different environmental conditions and object types is reduced
Solution Approach 1:
The patent implements dynamic adjustment of the region of interest based on detected motion patterns and environmental context. Rather than using fixed criteria, the system adapts the ROI boundaries and weighting parameters in real-time according to the observed scene characteristics, maintaining both computational efficiency and environmental adaptability.
Solution Approach 2:
The patent changes key parameters such as motion threshold, saliency weighting, and ROI expansion factors based on environmental conditions and detection confidence levels. These parameter adjustments allow the system to maintain fast computation while adapting to diverse environments and object types encountered during navigation.
3Loss of time
If ontology-based saliency is implemented to focus on predefined areas of interest, then regions of interest are narrowed down efficiently, but system complexity increases due to ontology integration
Solution Approach 1:
The patent performs preliminary action by pre-defining ontological categories and associated motion patterns before deployment. During operation, the system simply matches observed motion against these pre-established ontological templates, avoiding the need for complex real-time ontology reasoning and reducing both processing time and operational complexity.
Solution Approach 2:
The patent uses simplified copies or representations of ontological knowledge in the form of predefined motion signatures and object templates. Rather than implementing full ontological reasoning, the system uses these compact representations to efficiently guide saliency detection and ROI identification, reducing system complexity while maintaining effectiveness.
4Device complexity
If visual change detection is performed without ontology control, then system simplicity is maintained, but ability to detect meaningful changes in complex scenarios is reduced
Solution Approach 1:
The patent applies local quality by implementing ontology-based control only in specific critical regions or for specific object types where meaningful change detection is most important. Rather than applying complex ontology reasoning globally, the system focuses ontological guidance on key areas, maintaining overall system simplicity while improving precision where it matters most.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A system and method for identifying real time change in a scene of an unknown environment using an unmanned vehicle. In the context of unmanned vehicle navigation, it is critical to calculate the saliency map in real time and employ them in scene understanding. This will reduce the search space and ensure that the process is quicker. A domain specific ontology is created and a saliency model is developed. The saliency model detects key domain specific regions of interest in two consecutive images. The regions of interest is used for registration and change detection. The system is detecting the change by using visual saliency as an abstract feature that is developed in the environment. Probability of change is derived using the salient maps of the two images.