《一种基于频谱段能量的高分辨率遥感图像边缘特征检测方法》由会员分享，可在线阅读，更多相关《一种基于频谱段能量的高分辨率遥感图像边缘特征检测方法（6页珍藏版）》请在金锄头文库上搜索。
1、null null40null5nullnullnullVol.40,No.5null 2011M10Acta Geodaetica etCartographica Sinica Oct.,2011cI|:1001-1595(2011)05-0587-05B?sOq*mH+_ZE吴桂平,肖鹏峰,冯学智,王null 珂,黄秋燕2vS,2,210093A Method of Edge Feature Detection from High-resolution Remote Sensing ImagesBased on FrequencySpectrumZone EnergyWUGuiping,X
2、IAO Pengfeng,FENGXuezhi,WANGKe,HUANGQiuyanDepartment ofGeographical InformationSciences,NanjingUniversity,Nanjing 210093,ChinaAbstract:In viewofthe characteristicsofhigh-resolution remotely sensed imagery,a noveledge feature detectionalgorithmbased on frequencyspectrumzoneenergy isintroduced. Fromth
3、eperspective ofenergy ofimagesignal,the methods of sliding-windowFourier transformation and frequency energy sectional superposition are used todecompose signalinto different frequency spectrumzones. Then according to good noise suppression capacity ofDC(direct current) central frequency spectrumzon
4、e image,contribution degree ofhigher-order frequency spectrumzone energy to the image edge feature was further analysed. Finally, the edge detection results are obtainedthroughaggregating spectrumenergy coefficientwithineachwindowradiusoffrequency spectrumrange from1 to1.5 and quantizing image gray
5、between 0-255. The experimental results shows that this algorithm remains goodedge ofimage and hasgood capability ofedge detection in a specific directionaswell.Keywords: edge feature; feature detection; frequency spectrum zone energy; window Fourier transform; high-resolutionremotelysensed imageryK
6、null1:从图像信号的能量角度出发,探讨一种基于频谱段能量的高分辨率遥感图像边缘检测的方法b该方法采用滑动窗口傅里叶变换和频谱能量分段叠置的手段,将含有噪声的图像信号分解到不同的频谱段中,然后根据直流中心频谱段图像良好的噪声抑制特性,在此基础上进一步分析其高阶频谱段能量对图像边缘特征的贡献作用,将3null 3的滑动子窗口中频谱半径为1 1.5范围内的频谱能量系数进行求和叠置,并量化至0 255的灰度范围,得到最终的边缘检测图像b试验结果表明,基于频谱能量所检测出的图像边缘特征响应较为显著,视觉效果尚可,并且具有特定方向上的边缘检测能力b1oM:边缘特征;特征检测;频谱段能量;窗口傅里叶变换
7、;高分辨率遥感图像ms|:P237null null null nullDSM:A:国家863计划(2008AA12Z106); 国家自然科学基金(40801166;40771137)1 nullnullH+_*msaSMY#u4|5/,-X*S5B1-3bC,1*mH+_5vb004-5,mI#S=M,H#B=Z_M?pLCmH_,Robert0aSobel0aLaplace0aLOG0b7bWsOq*m,;+FA,7OmMa.2aHv+,ts$b_0sOq*m7,4rzH_rT6-7bmHLbMVC,HVYV%?V8,qT?+B,8H+9-10bHsv3g=MZEbl|+sOqs?11-1
8、2,mH_4mbN,1IKONOSsOq*m$,V)B?mH+_ZE,ZEn5sv=M,s?Va,YVs?mH+DT,HOctober 2011 Vol.40 No.5AGCS http: mz.2+,LCsOq*mH+r_b2 null?*mLb=f,csadsaaH#.2,7tsMB,b4|s7b=,BBHq=fVsB?f=FblBfVbWqM,=M13F(u, v) = 1MN nullM- 1x= 0 nullN- 1y= 0f (x , y)e- j2null uxM+ vyN (1)T(1)BjM null N=mVbWq=M,F(u, v )c$f ( x, y),qs|F( u
9、, v )|vlvQBmbWs14bm,B(Vbqs,7OvlsYQs|_bH,m1Q?s,mH%H,5mQ?9bNVn,=M/,V|msf (x, y)M?sF(u, v),PqsmzQ,V7ZLLCma.2#Hs4|15-16bParseval,=Mq|F( u, v)|2bmf (x, y)?iM1nullM- 1x= 0 nullN- 1y= 0f ( x, y) 2 = 1MN nullM- 1u= 0 nullN- 1v= 0F(u, v ) 2 (2)M-9?M,b?(x, y)s,7,?v(u, v)bb=m9?q=q?b,TL!vlM null NmsuC,suVsnM0C i(m1U),OC = C1+ C2+ null+ Cn,5null null f (x, y)Cx,y 2 = null null F(u,v)C1 2 + null+null null F(u, v) 2Ci(3)T,?null null | FCi(u, v)|2VUmq0Ci?vl,?,M0Ci17byN,a
