【摘要】 针对当前地震云研究中只能利用长期以来的经验进行目视解译,无法形成系统的研究理论和研究模型的现状,结合目前流行的Android便携式设备,提出在Android平台上对地震云图像进行轮廓提取与匹配的研究思路。采用抗噪性能较好的数学形态学算法,在此基础上得到一种改进算法以提取出感兴趣区域,进而通过构造10个新的不变矩对Hu矩匹配算法进行改进,进行轮廓匹配,识别出地震云图像。在Android平台上则利用NDK(Native Delelopment Kit)通过JNI(Java Native Interface)调用OpenCV库函数,用C++语言实现改进算法。实验结果表明:该方法能够有效地识别时间序列下的地震云,正确率高,并且在Android平台上操作简单方便,为地震云预测地震提供初步研究基础和研究依据。更多还原

【Abstract】 The research on pre-earthquake cloud is limited to visual interpretation,and it hasn’t form a unified scientific theory.Pointed to this situation,and combined with the popular portable Android platform,the idea of extracting and matching pre-earthquake contour based on Android platform is proposed.Firstly,an improved mathematical morphology is used to extractpre-earthquake cloud contour.And then,10 new invariant moments of original image and reference image are calculated by using an improved Hu moments algorithm.Finally,match results are obtained to recognize the pre-earthquake cloud.The algorithms is realized on Android platform,it uses C++language and calls OpenCV library functions through Native Development Kit(NDK)and Java Native Interface(JNI).The result shows the method can well realize pre-earthquake cloud recognition under time series.It is easy to operate on Android device,and can make foundation for deep research on preearthquake cloud.更多还原