Edge detection methods applied to the analysis of spherical raindrop images

Area of Science:

• Atmospheric science and meteorology
• Optical physics and instrumentation
• Computer vision and image processing

Background:

• Optical imaging of raindrops is vital for determining raindrop size and shape distributions.
• These distributions are essential for calculating rainfall rates from radar data.
• Automatic image processing is necessary due to the large volume of data required.

Purpose of the Study:

• To evaluate eight edge detection algorithms for raindrop image analysis.
• To determine the optimal algorithm for accurate raindrop size measurement.
• To identify an algorithm that maintains a large depth of field (dof) for improved statistics.

Main Methods:

• Comparison of eight different edge detection algorithms applied to raindrop images.
• Analysis of algorithm performance in terms of drop outline extraction accuracy.
• Assessment of the trade-off between measurement accuracy and depth of field.

Main Results:

• The accuracy of raindrop size measurement is highly dependent on the image processing algorithm.
• A larger depth of field is needed for comprehensive statistical analysis.
• The Hueckel algorithm demonstrated superior performance, offering the largest dof and most accurate size estimation.

Conclusions:

• The Hueckel algorithm is the most suitable for optical raindrop imaging systems requiring accurate size and shape statistics.
• This finding aids in improving rainfall rate estimations derived from radar measurements.
• The chosen algorithm effectively addresses the challenge of balancing measurement accuracy with depth of field.