Minimum-mean-square-error filters for detecting a noisy target in background noise

Area of Science:

• Signal processing
• Image analysis
• Filter design

Background:

• Detecting targets in noisy images is challenging.
• Existing methods struggle with spatially nonoverlapping background noise.

Purpose of the Study:

• To develop a minimum-mean-square-error (MMSE) filter for noisy target detection.
• To consider both target-additive noise and spatially nonoverlapping background noise.

Main Methods:

• Designed a filter minimizing mean-square-error between output and a target delta function.
• Evaluated filter performance using computer simulations on noisy images.

Main Results:

• The proposed MMSE filter successfully detected noisy targets.
• The filter showed good discrimination against undesired objects.
• The filter exhibited tolerance to target distortions like rotation and scaling.

Conclusions:

• The MMSE filter is a viable solution for detecting noisy targets in complex noise environments.
• The filter's robustness to distortions enhances its practical applicability.