Learned linear models for detecting watercraft in a maritime environment.

Area of Science:

• Computer Vision
• Signal Processing
• Remote Sensing

Background:

• Object detection in scenes is crucial for surveillance and monitoring.
• Wavelet models are established tools for signal analysis and feature extraction.
• Maritime target detection presents challenges due to environmental conditions and target variability.

Purpose of the Study:

• To develop a new linear model for object detection using generalized wavelets.
• To estimate model parameters for enhanced detection of maritime targets.
• To evaluate the proposed model's performance against standard wavelet techniques.

Main Methods:

• Utilizing representative training data to estimate generalized wavelet model parameters.
• Employing a maximum likelihood approach to optimize detection probability for a fixed false alarm rate.
• Testing the model on a short-wave infrared (SWIR) imagery database of watercraft.

Main Results:

• The proposed generalized wavelet model demonstrated effective detection of maritime targets.
• Parameter estimation maximized the probability of target detection.
• Performance comparison showed advantages over traditional wavelet bases in detection tasks.

Conclusions:

• The novel generalized wavelet model offers a promising approach for maritime object detection.
• The method provides a robust framework for analyzing infrared imagery.
• This technique enhances the capability for identifying vessels in complex marine environments.