Cloud precipitation particle shape recognition method based on image geometric-feature parameters

Abstract

The invention discloses a cloud precipitation particle shape recognition method based on image geometric-feature parameters, belongs to the field of digital-image recognition, and aims to recognize and determine, on the basis of automatically detecting a cloud precipitation particle target, shape of the cloud precipitation particle target to improve automatic-analysis capability of a cloud micro-physical-process. The method includes: (1) a cloud precipitation particle image acquisition step; (2) a cloud precipitation particle target detection and extraction step; (3) a cloud precipitation particle shape image database establishment step; (4) a cloud precipitation particle shape geometric-feature parameter extraction and selection step; and (5) a cloud precipitation particle shape recognition step. According to the cloud precipitation particle shape recognition method based on the image geometric-feature parameters of the invention, the geometric-feature parameters which can effectivelycharacterize the particle shape are selected through establishing a cloud precipitation particle shape image database, and accuracy and speed of recognition are improved through optimizing the recognition step. The method can be used for occasions of airborne cloud precipitation particle automatic-measurement.

Description

technical field [0001] The invention belongs to the field of digital image recognition, and in particular relates to a cloud precipitation particle shape recognition method based on image geometric feature parameters, which is used to automatically recognize the shape information of cloud precipitation particles, so as to improve the ability of automatic analysis of cloud microphysical processes. Background technique [0002] Using aircraft to carry observation instruments directly into the cloud to detect is an important observation method in the research of cloud precipitation physics and weather modification. At the same time, the data obtained from the observation of aircraft entering the cloud can also be used to verify the results of remote sensing measurements such as radar and satellite, as well as the results of model and parametric simulation. [0003] Identifying particle shapes in clouds is important for understanding changes in cloud microphysics. The cloud par...

AI Technical Summary

Problems solved by technology

[0004] In 1981, Rahman et al. of the University of Wyoming in the United States collected cloud particle images by aircraft, and used the adaptive Kalman filter method combined with Bayesian decision theory to divide the complete and partial cloud particle images into seven basic particle shapes (plate shape, star shape, columnar shape, space radial branch shape, hat columnar shape, graupel and raindrop), but this method has a large amount of calculation, which affects its real-time performance

[0005] In 1987, Holroyd of the U.S. Bureau of Land Reclamation developed a set of technology to identify and classify the cloud particle images collected by the airborne instrument 2DC by using image geometric feature parameters. shape, aggregation shape, graupel shape, spherical shape, plate shape, irregular shape and dendritic shape, this method is simple and easy to operate, and the operation speed is fast, but the recognition accuracy is limited

[0006] In 2000, Korolev and Sussman of the Canadian Meteorological Bureau used the acquired geometric characteristics of cloud particles, adopted a simple dimensionless ratio relationship, and solved the inverse problem to divide the particle shapes into spherical, irregular, needle-shaped and dendrite-shaped, but Korolev The method proposed by Sussman and Sussman is a probabilistic statistical identification method, which cannot identify individual particles

[0008] Although there have been many methods for automatic recognition of cloud and precipitation particle shapes in the past three decades, they all have certain limitations.

Images