Background modeling for moving object detection in long-distance imaging through turbulent medium.

Area of Science:

• Computer Vision
• Image Processing
• Remote Sensing

Background:

• Background modeling is essential for moving object detection.
• Long-distance imaging faces challenges from atmospheric turbulence, causing blur and movement.
• These factors complicate the temporal intensity distribution, especially at image edges.

Purpose of the Study:

• To theoretically and experimentally investigate background modeling for long-distance imaging.
• To determine appropriate statistical distributions for modeling background intensity.
• To analyze the impact of image blur and spatiotemporal movements on background models.

Main Methods:

• Theoretical analysis of background modeling under turbulent conditions.
• Experimental validation across various long-distance imaging scenarios.
• Evaluation of unimodal and multimodal distribution assumptions.

Main Results:

• A unimodal distribution is generally more suitable for background modeling in long-distance imaging.
• Image deblurring can alter the suitability of distribution models.
• Multimodal modeling may become more appropriate after image deblurring.

Conclusions:

• The choice of background model distribution depends on imaging conditions, particularly the presence of blur.
• Understanding intensity distribution is key to robust moving object detection in challenging environments.
• Deblurring techniques necessitate a re-evaluation of background modeling strategies.