Thursday, 17 September 2015
Oklahoma F (Embassy Suites Hotel and Conference Center )
The new generation weather radar is an observation platform with complex components. Due to various factors, including climate, hardware failure etc., abnormal radar echoes were brought out, resulting in inconvenience for forecasting and warning applications. Currently, there are two main methods for identifying abnormal radar echoes : base data based method and image feature based method. The base data based method screens the abnormal echoes mainly according to the physical characteristics of base data. The image feature based method attempts to explore adjacent relationship of pixels in image domain to detect the abnormal echoes, which is more intuitive than the base data based method.
In this paper, the abnormal radar echo images were firstly analyzed. In order to describe the image features of abnormal echoes with robustness to rotation, zoom, etc., log polar coordinates is exploited to represent echo images rather than Cartesian coordinates. In the log polar coordinates, the echo images were converted to simple rectangular areas, and the shape characteristics of abnormal echoes were represented in an easy way. Then the projection integral method is used to detect abnormal echo characteristics, and the identification of abnormal echo images is implemented according to the threshold of projection integral values. The experiment results show that the proposed method can effectively indentify common types of abnormal echo images in a simple and fast way.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner