基于单幅红外云图的有眼台风自动定位算法  

Algorithm of Eyed Typhoon Locating Based on Single Infrared Image

作　　者：张军[1] 刘正光[1] 吴冰[1] 刘还珠 

机构地区：[1]天津大学电气与自动化工程学院,天津300072 [2]国家气象中心,北京100081

出　　处：《天津大学学报（自然科学与工程技术版）》2005年第5期437-442,共6页Journal of Tianjin University：Science and Technology

基　　金：国家气象局"九五"重点攻关资助项目(BX0104)

摘　　要：基于单幅卫星红外通道图像,提出了一种智能与数学形态学相结合的有眼台风中心定位算法.算法首先给出有眼台风中心定位的规则,然后使用自定义模板去除灰度较高且灰度差相近的密闭云墙区域,再结合数学形态学理论,使用种子填充算法提取出候选眼区,最后根据台风眼区的特点,将灰度信息与距离信息结合,提出智能识别眼区的判句,经计算机自动分析推理,将使判决函数值最小的候选眼区确定为有眼台风中心.初步气象业务应用表明此算法有较高的识别率.An algorithm for the eyed typhoon center locating is proposed, it uses the mathematical morphology theory combining with intelligence method based on a single infrared-channel images of synchronous satellite. The locating rules for the eyed typhoon center are given in knowledge of meteorological field. Because the airtight cloud-wall area has high gray value and low gray value difference, it can be removed by using the defined template, and the candidate eye areas are extracted by using the seed-filling algorithm with mathematical morphology. According to the characteristics of the typhoon eye area, and combining the gray value information with the distance information, the decisional sentence for intelligently recognizing the eye area is structured. After the automatic calculation and illation of the computer, the candidate eye area whose corresponding decisional value is minimum is taken as the eyed typhoon center. The experiments prove that the algorithm has high locating precision in the meteorological applications, and can satisfy the weather forecast work.

关 键 词：有眼台风 数学形态学 中心定位 红外卫星云图 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]