level 6
渴望被苹果砸中
楼主
NVI二次开发代码共享之“监督与非监督分类”
;+
;ENVI二次开发功能代码
;
;Author: DYQ
;问题讨论：
; http://hi.baidu.com/dyqwrp
; http://bbs.esrichina-bj.cn/ESRI/?fromuid=9806
;描述：
; 分类处理，包括监督分类和非监督分类
;
;调用方法：
;
;CAL_CLASS,inputfile,outputfile, method,...
;
;inputFile:待分类影像
;outpurfile:分类结果;
;Method : 0--5为监督分类，6、7为非监督分类
;
; 1-最小距离 1
; 2-最大似然 2
; 3-马氏距离 5
; 4-神经元网络 ENVI_NEURAL_NET_DOIT
; 5-向量机 ENVI_SVM_DOIT
; 6-IsoData 4
; 7-K-Means 7
;注意：每一种算法需要使用的参数说明可参考ENVI帮助文档
;
PRO CAL_CLASS,inputfile,outputfile, method,$
;感兴趣区文件
roifile = roifile,$
;平行六面体分类算法可选参数
STDV = stdv, $
STD_MULT =STD_MULT,$
;神经元网络分类算法参数
theta = theta, $
eta = eta, $
alpha = alpha, $
act_type = act_type, $
rms_crit = rms_crit, $
num_layers = num_layers, $
num_sweeps = num_sweeps, $
;向量机
thresh = thresh, $
penalty = penalty, $
kernel_type = kernel_type, $
kernel_degree = kernel_degree, $
kernel_bias = kernel_bias ,$
;K-Means 算法可选参数
ITERATIONS = ITERATIONS, $
NUM_CLASSES = NUM_CLASSES , $
;ISO算法参数
CHANGE_THRESH = CHANGE_THRESH, $
ISO_MERGE_DIST = ISO_MERGE_DIST, $
ISO_MERGE_PAIRS = ISO_MERGE_PAIRS, $
ISO_MIN_PIXELS = ISO_MIN_PIXELS, $
ISO_SPLIT_SMULT = ISO_SPLIT_SMULT, $
ISO_SPLIT_STD = ISO_SPLIT_STD, $
MIN_CLASSES = MIN_CLASSES
COMPILE_OPT idl2
CATCH, Error_status
errorshow = 'Sorry to see the error,'+ $
' please send the error Information to "[email protected]"'
IF Error_status NE 0 THEN BEGIN
tmp = DIALOG_MESSAGE(errorshow+STRING(13b)+$
!ERROR_STATE.MSG,/error,title = '错误提示!')
return
ENDIF
;输入数据预处理
ENVI_OPEN_FILE, inputfile, r_fid=fid
IF (fid EQ -1) THEN BEGIN
RETURN
ENDIF
;获取文件信息
ENVI_FILE_QUERY, fid, dims=dims, nb=nb
pos = LINDGEN(nb)
out_name = outputfile
CASE method OF
;-IsoData 4
6: BEGIN
IF ~KEYWORD_SET(CHANGE_THRESH) THEN CHANGE_THRESH = .05
IF ~KEYWORD_SET(NUM_CLASSES) THEN NUM_CLASSES = 10
IF ~KEYWORD_SET(ITERATIONS) THEN ITERATIONS = 1
IF ~KEYWORD_SET(ISO_MERGE_DIST) THEN ISO_MERGE_DIST = 1
IF ~KEYWORD_SET(ISO_MERGE_PAIRS) THEN ISO_MERGE_PAIRS = 2
IF ~KEYWORD_SET(ISO_MIN_PIXELS) THEN ISO_MIN_PIXELS = 1
IF ~KEYWORD_SET(ISO_SPLIT_SMULT) THEN ISO_SPLIT_SMULT = 1
IF ~KEYWORD_SET(ISO_SPLIT_STD) THEN ISO_SPLIT_STD = 1
IF ~KEYWORD_SET(MIN_CLASSES) THEN MIN_CLASSES = 5
out_bname = 'IsoData'
ENVI_DOIT, 'class_doit', fid=fid, pos=pos, dims=dims, $
out_bname=out_bname, out_name=out_name, method=4, $
r_fid=r_fid, $
NUM_CLASSES = NUM_CLASSES, $
ITERATIONS = ITERATIONS, $
in_memory=0, $
CHANGE_THRESH = CHANGE_THRESH, $
ISO_MERGE_DIST = ISO_MERGE_DIST, $
ISO_MERGE_PAIRS = ISO_MERGE_PAIRS, $
ISO_MIN_PIXELS = ISO_MIN_PIXELS, $
ISO_SPLIT_SMULT = ISO_SPLIT_SMULT, $
ISO_SPLIT_STD = ISO_SPLIT_STD, $
MIN_CLASSES = MIN_CLASSES
END
;-K-Means 7
7: BEGIN
IF ~KEYWORD_SET(NUM_CLASSES) THEN NUM_CLASSES = 5
IF ~KEYWORD_SET(CHANGE_THRESH) THEN CHANGE_THRESH = .5
IF ~KEYWORD_SET(ITERATIONS) THEN ITERATIONS = 1
out_bname = 'K-Means'
thresh=REPLICATE(0.05,num_classes)
ENVI_DOIT, 'class_doit', fid=fid, pos=pos, dims=dims, $
out_bname=out_bname, out_name=out_name, method=7, $
r_fid=r_fid, $
lookup = BYTARR(3,num_classes+1), $
NUM_CLASSES = NUM_CLASSES, $
in_memory=0, CHANGE_THRESH=CHANGE_THRESH,$
ITERATIONS = ITERATIONS
END
;-平行六面体 0
0: BEGIN
ENVI_RESTORE_ROIS, roifile
roi_ids = ENVI_GET_ROI_IDS(fid=fid, $
roi_colors=roi_colors, roi_names=class_names)
class_names = ['Unclassified', class_names]
num_classes = N_ELEMENTS(roi_ids)
; Set the unclassified class to black and use roi colors
lookup = BYTARR(3,num_classes+1)
lookup[0,1] = roi_colors
; 计算类ROI的基本统计信息
mean = FLTARR(N_ELEMENTS(pos), num_classes)
stdv = FLTARR(N_ELEMENTS(pos), num_classes)
cov = FLTARR(N_ELEMENTS(pos),N_ELEMENTS(pos),num_classes)
FOR j=0, num_classes-1 DO BEGIN
;
roi_dims=[ENVI_GET_ROI_DIMS_PTR(roi_ids[j]),0,0,0,0]
ENVI_DOIT, 'envi_stats_doit', fid=fid, pos=pos, $
dims=roi_dims, comp_flag=4, mean=c_mean, $
stdv=c_stdv, cov=c_cov
MEAN[0,j] = c_mean
stdv[0,j] = c_stdv
cov[0,0,j] = c_cov
ENDFOR
;
thresh=REPLICATE(0.05,num_classes)
out_bname = 'parallelepiped'
ENVI_DOIT, 'class_doit', fid=fid, pos=pos, dims=dims, $
out_bname=out_bname, out_name=out_name, method=0, $
mean=mean, stdv=stdv, std_mult=st_mult, $
lookup=lookup, class_names=class_names, $
in_memory=0;, thresh=thresh
END
;-最小距离 1
1: BEGIN
ENVI_RESTORE_ROIS, roifile
roi_ids = ENVI_GET_ROI_IDS(fid=fid, $
roi_colors=roi_colors, roi_names=class_names)
class_names = ['Unclassified', class_names]
num_classes = N_ELEMENTS(roi_ids)
; Set the unclassified class to black and use roi colors
lookup = BYTARR(3,num_classes+1)
lookup[0,1] = roi_colors
; 计算类ROI的基本统计信息
;
mean = FLTARR(N_ELEMENTS(pos), num_classes)
stdv = FLTARR(N_ELEMENTS(pos), num_classes)
cov = FLTARR(N_ELEMENTS(pos),N_ELEMENTS(pos),num_classes)
FOR j=0, num_classes-1 DO BEGIN
;
roi_dims=[ENVI_GET_ROI_DIMS_PTR(roi_ids[j]),0,0,0,0]
ENVI_DOIT, 'envi_stats_doit', fid=fid, pos=pos, $
dims=roi_dims, comp_flag=4, mean=c_mean, $
stdv=c_stdv, cov=c_cov
MEAN[0,j] = c_mean
stdv[0,j] = c_stdv
cov[0,0,j] = c_cov
ENDFOR
;
thresh=REPLICATE(0.05,num_classes)
out_bname = 'MinimumDistance'
ENVI_DOIT, 'class_doit', fid=fid, pos=pos, dims=dims, $
out_bname=out_bname, out_name=out_name, method=1, $
mean=mean, stdv=stdv, std_mult=st_mult, $
lookup=lookup, class_names=class_names, $
in_memory=0
END
;-最大似然 2
2: BEGIN
ENVI_RESTORE_ROIS, roifile
roi_ids = ENVI_GET_ROI_IDS(fid=fid, $
roi_colors=roi_colors, roi_names=class_names)
class_names = ['Unclassified', class_names]
num_classes = N_ELEMENTS(roi_ids)
; Set the unclassified class to black and use roi colors
lookup = BYTARR(3,num_classes+1)
lookup[0,1] = roi_colors
; 计算类ROI的基本统计信息
;
mean = FLTARR(N_ELEMENTS(pos), num_classes)
stdv = FLTARR(N_ELEMENTS(pos), num_classes)
cov = FLTARR(N_ELEMENTS(pos),N_ELEMENTS(pos),num_classes)
FOR j=0, num_classes-1 DO BEGIN
;
roi_dims=[ENVI_GET_ROI_DIMS_PTR(roi_ids[j]),0,0,0,0]
ENVI_DOIT, 'envi_stats_doit', fid=fid, pos=pos, $
dims=roi_dims, comp_flag=4, mean=c_mean, $
stdv=c_stdv, cov=c_cov
MEAN[0,j] = c_mean
stdv[0,j] = c_stdv
cov[0,0,j] = c_cov
ENDFOR
;
thresh=REPLICATE(0.05,num_classes)
out_bname = 'MaximumLikelihood'
ENVI_DOIT, 'class_doit', fid=fid, pos=pos, dims=dims, $
out_bname=out_bname, out_name=out_name, method=2, $
mean=mean, stdv=stdv, std_mult=st_mult, $
lookup=lookup, class_names=class_names, $
cov = cov,$
in_memory=0
END
;-马氏距离 5
3: BEGIN
ENVI_RESTORE_ROIS, roifile
roi_ids = ENVI_GET_ROI_IDS(fid=fid, $
roi_colors=roi_colors, roi_names=class_names)
class_names = ['Unclassified', class_names]
num_classes = N_ELEMENTS(roi_ids)
; Set the unclassified class to black and use roi colors
lookup = BYTARR(3,num_classes+1)
lookup[0,1] = roi_colors
; 计算类ROI的基本统计信息
;
mean = FLTARR(N_ELEMENTS(pos), num_classes)
stdv = FLTARR(N_ELEMENTS(pos), num_classes)
cov = FLTARR(N_ELEMENTS(pos),N_ELEMENTS(pos),num_classes)
FOR j=0, num_classes-1 DO BEGIN
;
roi_dims=[ENVI_GET_ROI_DIMS_PTR(roi_ids[j]),0,0,0,0]
ENVI_DOIT, 'envi_stats_doit', fid=fid, pos=pos, $
dims=roi_dims, comp_flag=4, mean=c_mean, $
stdv=c_stdv, cov=c_cov
MEAN[0,j] = c_mean
stdv[0,j] = c_stdv
cov[0,0,j] = c_cov
ENDFOR
;
thresh=REPLICATE(0.05,num_classes)
out_bname = 'Mahalanobis'
ENVI_GET_ROI_INFORMATION, roi_ids,nPts = nPts
ENVI_DOIT, 'class_doit', fid=fid, pos=pos, dims=dims, $
out_bname=out_bname, out_name=out_name, method=5, $
mean=mean, stdv=stdv, std_mult=st_mult, $
lookup=lookup, class_names=class_names, $
cov = cov,NPTS = nPts, $
in_memory=0
END
;-神经元网络 ENVI_NEURAL_NET_DOIT
4: BEGIN
IF ~KEYWORD_SET(theta) THEN theta = .9
IF ~KEYWORD_SET(eta) THEN eta = .2
IF ~KEYWORD_SET(alpha) THEN alpha = .9
IF ~KEYWORD_SET(act_type) THEN act_type = 0
IF ~KEYWORD_SET(rms_crit) THEN rms_crit = .1
IF ~KEYWORD_SET(num_layers) THEN num_layers = 3
IF ~KEYWORD_SET(num_sweeps) THEN num_sweeps = 10
ENVI_RESTORE_ROIS, roifile
roi_ids = ENVI_GET_ROI_IDS(fid=fid, $
roi_colors=lookup, roi_names=class_names)
; Set the classification variables
;
num_classes = N_ELEMENTS(roi_ids)
class_names = ['Unclassified', class_names]
lookup = REFORM([0,0,0, $
REFORM(lookup,3*num_classes)],3,num_classes+1)
;
; Call the doit
;
ENVI_DOIT, 'envi_neural_net_doit', $
fid=fid, pos=pos, dims=dims, $
out_name=out_name, rule_out_name='', $
theta=theta, eta=eta, alpha=alpha, $
num_classes=num_classes, num_sweeps=num_sweeps, $
num_layers=num_layers, act_type=act_type, $
rms_crit=rms_crit, roi_ids=roi_ids, /train, $
class_names=class_names, lookup=lookup
END
;-向量机 ENVI_SVM_DOIT
5: BEGIN
IF ~KEYWORD_SET(thresh) THEN thresh = .5
IF ~KEYWORD_SET(penalty) THEN penalty=75
IF ~KEYWORD_SET(kernel_type) THEN kernel_type=1
IF ~KEYWORD_SET(kernel_degree) THEN kernel_degree=3
IF ~KEYWORD_SET(kernel_bias) THEN kernel_bias = 2.
ENVI_RESTORE_ROIS, roifile
roi_ids = ENVI_GET_ROI_IDS(fid=fid)
; Call the svm classification doit routine
envi_doit, 'envi_svm_doit', $
fid=fid, pos=pos, dims=dims, $
out_name=out_name, $
roi_ids=roi_ids, thresh=thresh, $
penalty=penalty, kernel_type= kernel_type, $
kernel_degree=kernel_degree, kernel_bias=kernel_bias
END
ELSE:
ENDCASE
END
2014年05月17日 06点05分
1
;+
;ENVI二次开发功能代码
;
;Author: DYQ
;问题讨论：
; http://hi.baidu.com/dyqwrp
; http://bbs.esrichina-bj.cn/ESRI/?fromuid=9806
;描述：
; 分类处理，包括监督分类和非监督分类
;
;调用方法：
;
;CAL_CLASS,inputfile,outputfile, method,...
;
;inputFile:待分类影像
;outpurfile:分类结果;
;Method : 0--5为监督分类，6、7为非监督分类
;
; 1-最小距离 1
; 2-最大似然 2
; 3-马氏距离 5
; 4-神经元网络 ENVI_NEURAL_NET_DOIT
; 5-向量机 ENVI_SVM_DOIT
; 6-IsoData 4
; 7-K-Means 7
;注意：每一种算法需要使用的参数说明可参考ENVI帮助文档
;
PRO CAL_CLASS,inputfile,outputfile, method,$
;感兴趣区文件
roifile = roifile,$
;平行六面体分类算法可选参数
STDV = stdv, $
STD_MULT =STD_MULT,$
;神经元网络分类算法参数
theta = theta, $
eta = eta, $
alpha = alpha, $
act_type = act_type, $
rms_crit = rms_crit, $
num_layers = num_layers, $
num_sweeps = num_sweeps, $
;向量机
thresh = thresh, $
penalty = penalty, $
kernel_type = kernel_type, $
kernel_degree = kernel_degree, $
kernel_bias = kernel_bias ,$
;K-Means 算法可选参数
ITERATIONS = ITERATIONS, $
NUM_CLASSES = NUM_CLASSES , $
;ISO算法参数
CHANGE_THRESH = CHANGE_THRESH, $
ISO_MERGE_DIST = ISO_MERGE_DIST, $
ISO_MERGE_PAIRS = ISO_MERGE_PAIRS, $
ISO_MIN_PIXELS = ISO_MIN_PIXELS, $
ISO_SPLIT_SMULT = ISO_SPLIT_SMULT, $
ISO_SPLIT_STD = ISO_SPLIT_STD, $
MIN_CLASSES = MIN_CLASSES
COMPILE_OPT idl2
CATCH, Error_status
errorshow = 'Sorry to see the error,'+ $
' please send the error Information to "[email protected]"'
IF Error_status NE 0 THEN BEGIN
tmp = DIALOG_MESSAGE(errorshow+STRING(13b)+$
!ERROR_STATE.MSG,/error,title = '错误提示!')
return
ENDIF
;输入数据预处理
ENVI_OPEN_FILE, inputfile, r_fid=fid
IF (fid EQ -1) THEN BEGIN
RETURN
ENDIF
;获取文件信息
ENVI_FILE_QUERY, fid, dims=dims, nb=nb
pos = LINDGEN(nb)
out_name = outputfile
CASE method OF
;-IsoData 4
6: BEGIN
IF ~KEYWORD_SET(CHANGE_THRESH) THEN CHANGE_THRESH = .05
IF ~KEYWORD_SET(NUM_CLASSES) THEN NUM_CLASSES = 10
IF ~KEYWORD_SET(ITERATIONS) THEN ITERATIONS = 1
IF ~KEYWORD_SET(ISO_MERGE_DIST) THEN ISO_MERGE_DIST = 1
IF ~KEYWORD_SET(ISO_MERGE_PAIRS) THEN ISO_MERGE_PAIRS = 2
IF ~KEYWORD_SET(ISO_MIN_PIXELS) THEN ISO_MIN_PIXELS = 1
IF ~KEYWORD_SET(ISO_SPLIT_SMULT) THEN ISO_SPLIT_SMULT = 1
IF ~KEYWORD_SET(ISO_SPLIT_STD) THEN ISO_SPLIT_STD = 1
IF ~KEYWORD_SET(MIN_CLASSES) THEN MIN_CLASSES = 5
out_bname = 'IsoData'
ENVI_DOIT, 'class_doit', fid=fid, pos=pos, dims=dims, $
out_bname=out_bname, out_name=out_name, method=4, $
r_fid=r_fid, $
NUM_CLASSES = NUM_CLASSES, $
ITERATIONS = ITERATIONS, $
in_memory=0, $
CHANGE_THRESH = CHANGE_THRESH, $
ISO_MERGE_DIST = ISO_MERGE_DIST, $
ISO_MERGE_PAIRS = ISO_MERGE_PAIRS, $
ISO_MIN_PIXELS = ISO_MIN_PIXELS, $
ISO_SPLIT_SMULT = ISO_SPLIT_SMULT, $
ISO_SPLIT_STD = ISO_SPLIT_STD, $
MIN_CLASSES = MIN_CLASSES
END
;-K-Means 7
7: BEGIN
IF ~KEYWORD_SET(NUM_CLASSES) THEN NUM_CLASSES = 5
IF ~KEYWORD_SET(CHANGE_THRESH) THEN CHANGE_THRESH = .5
IF ~KEYWORD_SET(ITERATIONS) THEN ITERATIONS = 1
out_bname = 'K-Means'
thresh=REPLICATE(0.05,num_classes)
ENVI_DOIT, 'class_doit', fid=fid, pos=pos, dims=dims, $
out_bname=out_bname, out_name=out_name, method=7, $
r_fid=r_fid, $
lookup = BYTARR(3,num_classes+1), $
NUM_CLASSES = NUM_CLASSES, $
in_memory=0, CHANGE_THRESH=CHANGE_THRESH,$
ITERATIONS = ITERATIONS
END
;-平行六面体 0
0: BEGIN
ENVI_RESTORE_ROIS, roifile
roi_ids = ENVI_GET_ROI_IDS(fid=fid, $
roi_colors=roi_colors, roi_names=class_names)
class_names = ['Unclassified', class_names]
num_classes = N_ELEMENTS(roi_ids)
; Set the unclassified class to black and use roi colors
lookup = BYTARR(3,num_classes+1)
lookup[0,1] = roi_colors
; 计算类ROI的基本统计信息
mean = FLTARR(N_ELEMENTS(pos), num_classes)
stdv = FLTARR(N_ELEMENTS(pos), num_classes)
cov = FLTARR(N_ELEMENTS(pos),N_ELEMENTS(pos),num_classes)
FOR j=0, num_classes-1 DO BEGIN
;
roi_dims=[ENVI_GET_ROI_DIMS_PTR(roi_ids[j]),0,0,0,0]
ENVI_DOIT, 'envi_stats_doit', fid=fid, pos=pos, $
dims=roi_dims, comp_flag=4, mean=c_mean, $
stdv=c_stdv, cov=c_cov
MEAN[0,j] = c_mean
stdv[0,j] = c_stdv
cov[0,0,j] = c_cov
ENDFOR
;
thresh=REPLICATE(0.05,num_classes)
out_bname = 'parallelepiped'
ENVI_DOIT, 'class_doit', fid=fid, pos=pos, dims=dims, $
out_bname=out_bname, out_name=out_name, method=0, $
mean=mean, stdv=stdv, std_mult=st_mult, $
lookup=lookup, class_names=class_names, $
in_memory=0;, thresh=thresh
END
;-最小距离 1
1: BEGIN
ENVI_RESTORE_ROIS, roifile
roi_ids = ENVI_GET_ROI_IDS(fid=fid, $
roi_colors=roi_colors, roi_names=class_names)
class_names = ['Unclassified', class_names]
num_classes = N_ELEMENTS(roi_ids)
; Set the unclassified class to black and use roi colors
lookup = BYTARR(3,num_classes+1)
lookup[0,1] = roi_colors
; 计算类ROI的基本统计信息
;
mean = FLTARR(N_ELEMENTS(pos), num_classes)
stdv = FLTARR(N_ELEMENTS(pos), num_classes)
cov = FLTARR(N_ELEMENTS(pos),N_ELEMENTS(pos),num_classes)
FOR j=0, num_classes-1 DO BEGIN
;
roi_dims=[ENVI_GET_ROI_DIMS_PTR(roi_ids[j]),0,0,0,0]
ENVI_DOIT, 'envi_stats_doit', fid=fid, pos=pos, $
dims=roi_dims, comp_flag=4, mean=c_mean, $
stdv=c_stdv, cov=c_cov
MEAN[0,j] = c_mean
stdv[0,j] = c_stdv
cov[0,0,j] = c_cov
ENDFOR
;
thresh=REPLICATE(0.05,num_classes)
out_bname = 'MinimumDistance'
ENVI_DOIT, 'class_doit', fid=fid, pos=pos, dims=dims, $
out_bname=out_bname, out_name=out_name, method=1, $
mean=mean, stdv=stdv, std_mult=st_mult, $
lookup=lookup, class_names=class_names, $
in_memory=0
END
;-最大似然 2
2: BEGIN
ENVI_RESTORE_ROIS, roifile
roi_ids = ENVI_GET_ROI_IDS(fid=fid, $
roi_colors=roi_colors, roi_names=class_names)
class_names = ['Unclassified', class_names]
num_classes = N_ELEMENTS(roi_ids)
; Set the unclassified class to black and use roi colors
lookup = BYTARR(3,num_classes+1)
lookup[0,1] = roi_colors
; 计算类ROI的基本统计信息
;
mean = FLTARR(N_ELEMENTS(pos), num_classes)
stdv = FLTARR(N_ELEMENTS(pos), num_classes)
cov = FLTARR(N_ELEMENTS(pos),N_ELEMENTS(pos),num_classes)
FOR j=0, num_classes-1 DO BEGIN
;
roi_dims=[ENVI_GET_ROI_DIMS_PTR(roi_ids[j]),0,0,0,0]
ENVI_DOIT, 'envi_stats_doit', fid=fid, pos=pos, $
dims=roi_dims, comp_flag=4, mean=c_mean, $
stdv=c_stdv, cov=c_cov
MEAN[0,j] = c_mean
stdv[0,j] = c_stdv
cov[0,0,j] = c_cov
ENDFOR
;
thresh=REPLICATE(0.05,num_classes)
out_bname = 'MaximumLikelihood'
ENVI_DOIT, 'class_doit', fid=fid, pos=pos, dims=dims, $
out_bname=out_bname, out_name=out_name, method=2, $
mean=mean, stdv=stdv, std_mult=st_mult, $
lookup=lookup, class_names=class_names, $
cov = cov,$
in_memory=0
END
;-马氏距离 5
3: BEGIN
ENVI_RESTORE_ROIS, roifile
roi_ids = ENVI_GET_ROI_IDS(fid=fid, $
roi_colors=roi_colors, roi_names=class_names)
class_names = ['Unclassified', class_names]
num_classes = N_ELEMENTS(roi_ids)
; Set the unclassified class to black and use roi colors
lookup = BYTARR(3,num_classes+1)
lookup[0,1] = roi_colors
; 计算类ROI的基本统计信息
;
mean = FLTARR(N_ELEMENTS(pos), num_classes)
stdv = FLTARR(N_ELEMENTS(pos), num_classes)
cov = FLTARR(N_ELEMENTS(pos),N_ELEMENTS(pos),num_classes)
FOR j=0, num_classes-1 DO BEGIN
;
roi_dims=[ENVI_GET_ROI_DIMS_PTR(roi_ids[j]),0,0,0,0]
ENVI_DOIT, 'envi_stats_doit', fid=fid, pos=pos, $
dims=roi_dims, comp_flag=4, mean=c_mean, $
stdv=c_stdv, cov=c_cov
MEAN[0,j] = c_mean
stdv[0,j] = c_stdv
cov[0,0,j] = c_cov
ENDFOR
;
thresh=REPLICATE(0.05,num_classes)
out_bname = 'Mahalanobis'
ENVI_GET_ROI_INFORMATION, roi_ids,nPts = nPts
ENVI_DOIT, 'class_doit', fid=fid, pos=pos, dims=dims, $
out_bname=out_bname, out_name=out_name, method=5, $
mean=mean, stdv=stdv, std_mult=st_mult, $
lookup=lookup, class_names=class_names, $
cov = cov,NPTS = nPts, $
in_memory=0
END
;-神经元网络 ENVI_NEURAL_NET_DOIT
4: BEGIN
IF ~KEYWORD_SET(theta) THEN theta = .9
IF ~KEYWORD_SET(eta) THEN eta = .2
IF ~KEYWORD_SET(alpha) THEN alpha = .9
IF ~KEYWORD_SET(act_type) THEN act_type = 0
IF ~KEYWORD_SET(rms_crit) THEN rms_crit = .1
IF ~KEYWORD_SET(num_layers) THEN num_layers = 3
IF ~KEYWORD_SET(num_sweeps) THEN num_sweeps = 10
ENVI_RESTORE_ROIS, roifile
roi_ids = ENVI_GET_ROI_IDS(fid=fid, $
roi_colors=lookup, roi_names=class_names)
; Set the classification variables
;
num_classes = N_ELEMENTS(roi_ids)
class_names = ['Unclassified', class_names]
lookup = REFORM([0,0,0, $
REFORM(lookup,3*num_classes)],3,num_classes+1)
;
; Call the doit
;
ENVI_DOIT, 'envi_neural_net_doit', $
fid=fid, pos=pos, dims=dims, $
out_name=out_name, rule_out_name='', $
theta=theta, eta=eta, alpha=alpha, $
num_classes=num_classes, num_sweeps=num_sweeps, $
num_layers=num_layers, act_type=act_type, $
rms_crit=rms_crit, roi_ids=roi_ids, /train, $
class_names=class_names, lookup=lookup
END
;-向量机 ENVI_SVM_DOIT
5: BEGIN
IF ~KEYWORD_SET(thresh) THEN thresh = .5
IF ~KEYWORD_SET(penalty) THEN penalty=75
IF ~KEYWORD_SET(kernel_type) THEN kernel_type=1
IF ~KEYWORD_SET(kernel_degree) THEN kernel_degree=3
IF ~KEYWORD_SET(kernel_bias) THEN kernel_bias = 2.
ENVI_RESTORE_ROIS, roifile
roi_ids = ENVI_GET_ROI_IDS(fid=fid)
; Call the svm classification doit routine
envi_doit, 'envi_svm_doit', $
fid=fid, pos=pos, dims=dims, $
out_name=out_name, $
roi_ids=roi_ids, thresh=thresh, $
penalty=penalty, kernel_type= kernel_type, $
kernel_degree=kernel_degree, kernel_bias=kernel_bias
END
ELSE:
ENDCASE
END
