Method and System for Automatic Object Detection and Subsequent Object Tracking According to Object Shape

Simple SummaryContent extracted from patent full text and abstract with AI.

This invention describes a method and system for automatically detecting and subsequently tracking objects in digital video systems. It uniquely combines a Gaussian Mixture Model (GMM)-based object detection algorithm with an enhanced Mean-Shift object tracking algorithm. The system not only identifies and tracks moving objects but can also determine the shape, contour, and three-dimensional orientation of objects. Key aspects include advanced shadow removal, generation of an asymmetric filter kernel based on object masks, and shape-adaptive tracking that makes the system robust even under complex, real-world video scenarios.

Use CasesContent extracted from patent full text and abstract with AI.

• Video surveillance and security monitoring systems (e.g., airports, public spaces, building entrances)
• Traffic management and monitoring (tracking vehicles, pedestrians)
• Autonomous vehicles and robotics (detecting and understanding the position and orientation of objects in real-time)
• Sports analytics (tracking players or equipment for performance analysis)
• Industrial automation and quality control (tracking and verifying objects on assembly lines)
• Augmented reality applications (accurate object recognition and pose estimation)
• Retail analytics (customer movement, interaction with products)

BenefitsContent extracted from patent full text and abstract with AI.

• Accurate detection and tracking of objects, including their shape and orientation in three-dimensional space, not just simple position tracking.
• Improved robustness against shadows and environmental changes due to advanced shadow removal techniques.
• Reduced false positive rates through the use of object masks and spatial-temporal correlations.
• Works without the need for extensive training data, making deployment easier and faster.
• Enhanced capability to track objects even after periods of occlusion or partial visibility.
• Efficiency and real-time applicability due to the inherent low computational complexity of the Mean-Shift algorithm.
• Enables early detection of dangerous or unusual object behavior (e.g., wrong-way movement on airport runways or roads).

Technical Classifications (CPCs)

Inventors & Applicants

Patent Abstract

A large number of tracking methods, in addition to further developments for object tracking, including tracking the object shape, are known. In this case, however, the orientation is calculated only within the image plane and therefore only 2-dimensionally, so the object shape cannot be matched to the actual movement of the object in three-dimensional space. In order to specify a method for automatic object detection and subsequent object tracking in digital video systems having at least one camera (K) for recording and transmitting video sequences, as well as further processing or evaluation of video data, which in addition to the localization and tracking of moving objects, is also capable of determining the object shape and thus of permitting the calculation of the orientation of an object in space with the existing measured data, patent claim 1 provides for an object detection algorithm based on a Gaussian mixture model GMM and expanded object tracking based on Mean-Shift to be combined with each other in the object detection. The object detection is expanded in accordance with a model of the background by improved removal of shadows, the binary mask (BM) generated in this way is used to create an asymmetric filter core, and then the actual algorithm for the shape-adaptive object tracking, expanded by a segmentation step for adapting the shape, is initialized, and therefore a determination at least of the object shape or object contour or the orientation of the object in space is made possible. The invention falls in the area of automatic object detection and subsequent object tracking.