Method and system of tracking an object based on multiple histograms

A precise object tracking system that excels even in low-quality images

This technology relates to an object tracking method and object tracking system. It relates to an object tracking method and device that is robust to the environment by obtaining multiple histograms for at least two features of a target object, selecting the one feature with the maximum cost among them, and applying a particle filter to track the object.

To overcome these limitations, this technology presents a particle filter-based object tracking technique that combines multiple histograms and Minimax estimation techniques. Accurate tracking is possible in any environment by analyzing two or more feature information, selecting the optimal feature, and calculating the particle weight.

‍

‍

‍

Key Features:

• A new object tracking method that combines the Minimax estimation technique with the Monte Carlo sampling (particle filter) method
• Simultaneously generates multiple histograms for multiple features such as the color (RGB, HSV) and shape (HOG) of the object
• Calculate the Bhattacharya distance with the reference model to obtain the cost, and select the one feature with the largest cost (maximum distance) among them
• Track lighting, shade, brightness changes, etc. through the Minimax technique When an interfering factor occurs, maintain stable tracking performance by using the feature most affected by the factor as a standard

‍