def save_fine_data(self, modeCfg):
"""
第二轮匹配，根据各通道的的mean和std计以为，角度和距离等进行精细化过滤
"""
# 最终的公共匹配点数量
idx = np.where(self.PubIdx > 0)
if len(idx[0]) == 0:
return
print u'所有粗匹配点数目 ', len(idx[0])
# 掩码清零
self.MaskRough[:] = 0
# 计算共同区域的距离差 #########
disDiff = np.full_like(self.S1_Time, '-1', dtype='i2')
a = np.power(self.S2_Lon[idx] - self.S1_Lon[idx], 2)
b = np.power(self.S2_Lat[idx] - self.S1_Lat[idx], 2)
disDiff[idx] = np.sqrt(a + b) * 100.
idx_Rough = np.logical_and(disDiff < modeCfg.distdif_max, disDiff >= 0)
idx1 = np.where(idx_Rough)
print u'1. 距离过滤后剩余点 ', len(idx1[0])
timeDiff = np.abs(self.S1_Time - self.S2_Time)
idx_Rough = np.logical_and(idx_Rough, timeDiff <= modeCfg.timedif_max)
idx1 = np.where(idx_Rough)
print u'2. 时间过滤后剩余点 ', len(idx1[0])
# 过滤太阳天顶角 ###############
idx_Rough = np.logical_and(
idx_Rough, self.S1_SoZ <= modeCfg.solzenith_max)
idx_Rough = np.logical_and(
idx_Rough, self.S2_SoZ <= modeCfg.solzenith_max)
idx1 = np.where(idx_Rough)
print u'3. 太阳天顶角过滤后剩余点 ', len(idx1[0])
# 计算耀斑角 ###############
glint1 = np.full_like(self.S1_SatZ, -999.)
glint2 = np.full_like(self.S1_SatZ, -999.)
glint1[idx] = sun_glint_cal(
self.S1_SatA[idx], self.S1_SatZ[idx], self.S1_SoA[idx], self.S1_SoZ[idx])
glint2[idx] = sun_glint_cal(
self.S2_SatA[idx], self.S2_SatZ[idx], self.S2_SoA[idx], self.S2_SoZ[idx])
idx_Rough = np.logical_and(idx_Rough, glint1 > modeCfg.solglint_min)
idx_Rough = np.logical_and(idx_Rough, glint2 > modeCfg.solglint_min)
print np.nanmin(glint1[idx]), np.nanmax(glint1[idx])
print np.nanmin(glint2[idx]), np.nanmax(glint2[idx])
idx1 = np.where(idx_Rough)
print u'4. 太阳耀斑角过滤后剩余点 ', len(idx1[0])
# 角度均匀性 #################
SatZRaio = np.full_like(self.S1_Time, 9999)
SatZ1 = np.cos(self.S1_SatZ[idx] * np.pi / 180.)
SatZ2 = np.cos(self.S2_SatZ[idx] * np.pi / 180.)
SatZRaio[idx] = np.abs(SatZ1 / SatZ2 - 1.)
idx_Rough = np.logical_and(idx_Rough, SatZRaio <= modeCfg.angledif_max)
idx1 = np.where(idx_Rough)
print u'5. 卫星天顶角均匀性过滤后剩余点 ', len(idx1[0])
idx_Rough = np.logical_and(
idx_Rough, self.S1_SatZ <= modeCfg.satzenith_max)
idx1 = np.where(idx_Rough)
print u'6. FY卫星观测角(天顶角)滤后剩余点 ', len(idx1[0])
self.MaskRough[idx1] = 1
for Band1 in modeCfg.chan1:
# 掩码清零
self.MaskFine[Band1][:] = 0
th_vaue_max = modeCfg.CH_threshold[Band1]['value_max']
th1 = modeCfg.CH_threshold[Band1]['angledif_max']
th2 = modeCfg.CH_threshold[Band1]['homodif_fov_max']
th3 = modeCfg.CH_threshold[Band1]['homodif_env_max']
th4 = modeCfg.CH_threshold[Band1]['homodif_fov_env_max']
flag = 0
# 可见光通道
if hasattr(self, 'S1_FovRefMean') and Band1 in self.S1_FovRefMean.keys():
flag = 'vis'
homoFov1 = np.abs(
self.S1_FovRefStd[Band1] / self.S1_FovRefMean[Band1])
homoEnv1 = np.abs(
self.S1_EnvRefStd[Band1] / self.S1_EnvRefMean[Band1])
homoFovEnv1 = np.abs(
self.S1_FovRefMean[Band1] / self.S1_EnvRefMean[Band1] - 1)
homoValue1 = self.S1_FovRefMean[Band1]
homoFov2 = np.abs(
self.S2_FovRefStd[Band1] / self.S2_FovRefMean[Band1])
homoEnv2 = np.abs(
self.S2_EnvRefStd[Band1] / self.S2_EnvRefMean[Band1])
homoFovEnv2 = np.abs(
self.S2_FovRefMean[Band1] / self.S2_EnvRefMean[Band1] - 1)
homoValue2 = self.S2_FovRefMean[Band1]
condition = np.logical_and(self.MaskRough > 0, True)
condition = np.logical_and(SatZRaio < th1, condition)
idx = np.where(condition)
print u'%s %s 角度均匀性过滤后，精匹配点个数 %d' % (Band1, flag, len(idx[0]))
condition = np.logical_and(homoValue1 < th_vaue_max, condition)
condition = np.logical_and(homoValue1 > 0, condition)
idx = np.where(condition)
print u'%s %s 饱和值过滤后，精匹配点个数 %d' % (Band1, flag, len(idx[0]))
condition = np.logical_and(homoFov1 < th2, condition)
idx = np.where(condition)
print u'%s %s 靶区过滤后，精匹配点个数 %d' % (Band1, flag, len(idx[0]))
condition = np.logical_and(homoEnv1 < th3, condition)
idx = np.where(condition)
print u'%s %s 环境过滤后，精匹配点个数 %d' % (Band1, flag, len(idx[0]))
condition = np.logical_and(homoFovEnv1 < th4, condition)
idx = np.where(condition)
print u'%s %s 靶区环境过滤后，精匹配点个数 %d' % (Band1, flag, len(idx[0]))
# sat 2过滤
condition = np.logical_and(homoValue2 > 0, condition)
condition = np.logical_and(homoValue2 < th_vaue_max, condition)
idx = np.where(condition)
print u'%s %s 饱和值2过滤后，精匹配点个数 %d' % (Band1, flag, len(idx[0]))
condition = np.logical_and(homoFov2 < th2, condition)
idx = np.where(condition)
print u'%s %s 靶区2过滤后，精匹配点个数 %d' % (Band1, flag, len(idx[0]))
condition = np.logical_and(homoEnv2 < th3, condition)
idx = np.where(condition)
print u'%s %s 环境2过滤后，精匹配点个数 %d' % (Band1, flag, len(idx[0]))
condition = np.logical_and(homoFovEnv2 < th4, condition)
idx = np.where(condition)
print u'%s %s 靶区环境2过滤后，精匹配点个数 %d' % (Band1, flag, len(idx[0]))
self.MaskFine[Band1][idx] = 1
def save_fine_data(self, modeCfg):
"""
第二轮匹配，根据各通道的的mean和std计以为，角度和距离等进行精细化过滤
"""
# 最终的公共匹配点数量
idx = np.where(self.PubIdx > 0)
if len(idx[0]) == 0:
return
print u'所有粗匹配点数目 ', len(idx[0])
# 掩码清零
self.MaskRough[:] = 0
# 计算共同区域的距离差 #########
disDiff = np.full_like(self.S1_Time, '-1', dtype='i2')
a = np.power(self.S2_Lon[idx] - self.S1_Lon[idx], 2)
b = np.power(self.S2_Lat[idx] - self.S1_Lat[idx], 2)
disDiff[idx] = np.sqrt(a + b) * 100.
idx_Rough = np.logical_and(disDiff < modeCfg.distdif_max, disDiff >= 0)
idx1 = np.where(idx_Rough)
print u'1. 距离过滤后剩余点 ', len(idx1[0])
timeDiff = np.abs(self.S1_Time - self.S2_Time)
idx_Rough = np.logical_and(idx_Rough, timeDiff <= modeCfg.timedif_max)
idx1 = np.where(idx_Rough)
print u'2. 时间过滤后剩余点 ', len(idx1[0])
# 过滤太阳天顶角 ###############
idx_Rough = np.logical_and(idx_Rough,
self.S1_SoZ <= modeCfg.solzenith_max)
idx_Rough = np.logical_and(idx_Rough,
self.S2_SoZ <= modeCfg.solzenith_max)
idx1 = np.where(idx_Rough)
print u'3. 太阳天顶角过滤后剩余点 ', len(idx1[0])
# 计算耀斑角 ###############
glint1 = np.full_like(self.S1_SatZ, -999.)
glint2 = np.full_like(self.S1_SatZ, -999.)
glint1[idx] = sun_glint_cal(self.S1_SatA[idx], self.S1_SatZ[idx],
self.S1_SoA[idx], self.S1_SoZ[idx])
glint2[idx] = sun_glint_cal(self.S2_SatA[idx], self.S2_SatZ[idx],
self.S2_SoA[idx], self.S2_SoZ[idx])
idx_Rough = np.logical_and(idx_Rough, glint1 > modeCfg.solglint_min)
idx_Rough = np.logical_and(idx_Rough, glint2 > modeCfg.solglint_min)
idx1 = np.where(idx_Rough)
print u'4. 太阳耀斑角过滤后剩余点 ', len(idx1[0])
# 角度均匀性 #################
SatZRaio = np.full_like(self.S1_Time, 9999)
SatZ1 = np.cos(self.S1_SatZ[idx] * np.pi / 180.)
SatZ2 = np.cos(self.S2_SatZ[idx] * np.pi / 180.)
SatZRaio[idx] = np.abs(SatZ1 / SatZ2 - 1.)
idx_Rough = np.logical_and(idx_Rough, SatZRaio <= modeCfg.angledif_max)
idx1 = np.where(idx_Rough)
print u'5. 卫星天顶角均匀性过滤后剩余点 ', len(idx1[0])
idx_Rough = np.logical_and(idx_Rough,
self.S1_SatZ <= modeCfg.satzenith_max)
idx1 = np.where(idx_Rough)
print u'6. FY卫星观测角(天顶角)滤后剩余点 ', len(idx1[0])
self.MaskRough[idx1] = 1
# 添加spec, 粗匹配后剩余的点是要记录光谱谱线的。。。2维转1维下标
idx_1d = np.ravel_multi_index(idx1, (self.row, self.col))
if modeCfg.write_spec:
# 定义spec_MaskRough_value 然后记录需要保存的谱线
if hasattr(self, 'S2_Spec_all'):
for i in idx_1d:
self.S2_Spectral.append(self.S2_Spec_all[i])
self.S2_Spectral = np.array(self.S2_Spectral)
# 记录根据MaskRough表记录的格点信息
self.S2_Spec_row = idx1[0]
self.S2_Spec_col = idx1[1]
for Band1 in modeCfg.chan1:
# 掩码清零
self.MaskFine[Band1][:] = 0
th_vaue_max = modeCfg.CH_threshold[Band1]['value_max']
th1 = modeCfg.CH_threshold[Band1]['angledif_max']
th2 = modeCfg.CH_threshold[Band1]['homodif_fov_max']
th3 = modeCfg.CH_threshold[Band1]['homodif_env_max']
th4 = modeCfg.CH_threshold[Band1]['homodif_fov_env_max']
th_cld1 = modeCfg.CH_threshold[Band1]['cld_angledif_max']
th_cld2 = modeCfg.CH_threshold[Band1]['cld_homodif_fov_max']
th_cld3 = modeCfg.CH_threshold[Band1]['cld_homodif_env_max']
th_cld4 = modeCfg.CH_threshold[Band1]['cld_homodif_fov_env_max']
flag = 0
# 如果 rad和tbb都有就用rad 做均匀性判断
if hasattr(self,
'S1_FovRadMean') and Band1 in self.S1_FovRadMean.keys():
flag = 'ir'
# 固定通道值用于检测红外晴空和云
# irValue = self.FovTbbMean1[modeCfg.clear_band_ir]
homoFov1 = np.abs(self.S1_FovRadStd[Band1] /
self.S1_FovRadMean[Band1])
homoEnv1 = np.abs(self.S1_EnvRadStd[Band1] /
self.S1_EnvRadMean[Band1])
homoFovEnv1 = np.abs(self.S1_FovRadMean[Band1] /
self.S1_EnvRadMean[Band1] - 1)
homoValue1 = self.S1_FovRadMean[Band1]
homoFov2 = np.abs(self.S2_FovRadStd[Band1] /
self.S2_FovRadMean[Band1])
homoEnv2 = np.abs(self.S2_EnvRadStd[Band1] /
self.S2_EnvRadMean[Band1])
homoFovEnv2 = np.abs(self.S2_FovRadMean[Band1] /
self.S2_EnvRadMean[Band1] - 1)
homoValue2 = self.S2_FovRadMean[Band1]
# 如果只有 tbb 就用tbb
if hasattr(self,
'S1_FovTbbMean') and not hasattr(self, 'S1_FovRadMean'):
if Band1 in self.S1_FovTbbMean.keys():
flag = 'ir'
# 固定通道值用于检测红外晴空和云
# irValue = self.FovTbbMean1[modeCfg.clear_band_ir]
homoFov1 = np.abs(self.FovTbbStd1[Band1] /
self.FovTbbMean1[Band1])
homoEnv1 = np.abs(self.EnvTbbStd1[Band1] /
self.EnvTbbMean1[Band1])
homoFovEnv1 = np.abs(self.FovTbbMean1[Band1] /
self.EnvTbbMean1[Band1] - 1)
homoValue1 = self.FovTbbMean1[Band1]
homoFov2 = np.abs(self.FovTbbStd2[Band1] /
self.FovTbbMean2[Band1])
homoEnv2 = np.abs(self.EnvTbbStd2[Band1] /
self.EnvTbbMean2[Band1])
homoFovEnv2 = np.abs(self.FovTbbMean2[Band1] /
self.EnvTbbMean2[Band1] - 1)
homoValue2 = self.FovTbbMean2[Band1]
# 可见光通道
if hasattr(self,
'S1_FovRefMean') and Band1 in self.S1_FovRefMean.keys():
flag = 'vis'
# visValue = self.FovRefMean1[modeCfg.clear_band_vis]
homoFov1 = np.abs(self.S1_FovRefStd[Band1] /
self.S1_FovRefMean[Band1])
homoEnv1 = np.abs(self.S1_EnvRefStd[Band1] /
self.S1_EnvRefMean[Band1])
homoFovEnv1 = np.abs(self.S1_FovRefMean[Band1] /
self.S1_EnvRefMean[Band1] - 1)
homoValue1 = self.S1_FovRefMean[Band1]
homoFov2 = np.abs(self.S2_FovRefStd[Band1] /
self.S2_FovRefMean[Band1])
homoEnv2 = np.abs(self.S2_EnvRefStd[Band1] /
self.S2_EnvRefMean[Band1])
homoFovEnv2 = np.abs(self.S2_FovRefMean[Band1] /
self.S2_EnvRefMean[Band1] - 1)
homoValue2 = self.S2_FovRefMean[Band1]
# 云判识关闭状态 ####
if (modeCfg.clear_min_ir == 0
and 'ir' in flag) or (modeCfg.clear_max_vis == 0
and 'vis' in flag):
condition = np.logical_and(self.MaskRough > 0, True)
condition = np.logical_and(SatZRaio < th1, condition)
idx = np.where(condition)
print u'%s %s 云判识关闭,角度均匀性过滤后，精匹配点个数 %d' % (Band1, flag,
len(idx[0]))
condition = np.logical_and(homoFov1 < th2, condition)
idx = np.where(condition)
print u'%s %s 云判识关闭,靶区过滤后，精匹配点个数 %d' % (Band1, flag, len(
idx[0]))
condition = np.logical_and(homoEnv1 < th3, condition)
idx = np.where(condition)
print u'%s %s 云判识关闭,环境过滤后，精匹配点个数 %d' % (Band1, flag, len(
idx[0]))
condition = np.logical_and(homoFovEnv1 < th4, condition)
idx = np.where(condition)
print u'%s %s 云判识关闭,靶区环境过滤后，精匹配点个数 %d' % (Band1, flag,
len(idx[0]))
condition = np.logical_and(homoValue1 < th_vaue_max, condition)
condition = np.logical_and(homoValue1 > 0, condition)
idx = np.where(condition)
print u'%s %s 云判识关闭,饱和值过滤后，精匹配点个数 %d' % (Band1, flag,
len(idx[0]))
# sat 2过滤
condition = np.logical_and(homoFov2 < th2, condition)
idx = np.where(condition)
print u'%s %s 云判识关闭,靶区2过滤后，精匹配点个数 %d' % (Band1, flag,
len(idx[0]))
condition = np.logical_and(homoEnv2 < th3, condition)
idx = np.where(condition)
print u'%s %s 云判识关闭,环境2过滤后，精匹配点个数 %d' % (Band1, flag,
len(idx[0]))
condition = np.logical_and(homoFovEnv2 < th4, condition)
idx = np.where(condition)
print u'%s %s 云判识关闭,靶区环境2过滤后，精匹配点个数 %d' % (Band1, flag,
len(idx[0]))
condition = np.logical_and(homoValue2 > 0, condition)
condition = np.logical_and(homoValue2 < th_vaue_max, condition)
idx = np.where(condition)
print u'%s %s 云判识关闭,饱和值2过滤后，精匹配点个数 %d' % (Band1, flag,
len(idx[0]))
self.MaskFine[Band1][idx] = 1
# 云判识开启状态 ####
else:
# 晴空判别
if 'ir' in flag:
# 固定通道值用于检测可见晴空和云
irValue = self.S1_FovTbbMean[modeCfg.clear_band_ir]
condition = np.logical_and(self.MaskRough > 0,
irValue >= modeCfg.clear_min_ir)
elif 'vis' in flag:
# 固定通道值用于检测可见晴空和云
visValue = self.S1_FovRefMean[modeCfg.clear_band_vis]
condition = np.logical_and(
self.MaskRough > 0, visValue < modeCfg.clear_max_vis)
condition = np.logical_and(visValue > 0, condition)
idx = np.where(condition)
test_idx = np.where(self.MaskRough > 0)
# print len(test_idx[0])
# print modeCfg.clear_band_ir, modeCfg.clear_min_ir
# print np.min(irValue[test_idx]), np.max(irValue[test_idx])
print u'%s %s 云判识开启,晴空点个数 %d' % (Band1, flag, len(idx[0]))
condition = np.logical_and(SatZRaio < th1, condition)
idx = np.where(condition)
print u'%s %s 云判识开启,晴空点中 角度 均匀过滤后, 精匹配点个数 %d' % (Band1, flag,
len(idx[0]))
condition = np.logical_and(homoFov1 < th2, condition)
idx = np.where(condition)
print u'%s %s 云判识开启,晴空点中 靶区1 均匀过滤后, 精匹配点个数 %d' % (Band1, flag,
len(idx[0]))
condition = np.logical_and(homoEnv1 < th3, condition)
idx = np.where(condition)
print u'%s %s 云判识开启,晴空点中 环境1 均匀过滤后, 精匹配点个数 %d' % (Band1, flag,
len(idx[0]))
condition = np.logical_and(homoFovEnv1 < th4, condition)
idx = np.where(condition)
print u'%s %s 云判识开启,晴空点中 靶区/环境 1 均匀过滤后, 精匹配点个数 %d' % (
Band1, flag, len(idx[0]))
condition = np.logical_and(homoValue1 > 0, condition)
condition = np.logical_and(homoValue1 < th_vaue_max, condition)
idx = np.where(condition)
print u'%s %s 云判识开启,晴空点中 饱和值1 均匀过滤后, 精匹配点个数 %d' % (Band1, flag,
len(idx[0]))
condition = np.logical_and(homoFov2 < th2, condition)
idx = np.where(condition)
print u'%s %s 云判识开启,晴空点中 靶区2 均匀过滤后, 精匹配点个数 %d' % (Band1, flag,
len(idx[0]))
condition = np.logical_and(homoEnv2 < th3, condition)
idx = np.where(condition)
print u'%s %s 云判识开启,晴空点中 环境2 均匀过滤后, 精匹配点个数 %d' % (Band1, flag,
len(idx[0]))
condition = np.logical_and(homoFovEnv2 < th4, condition)
idx = np.where(condition)
print u'%s %s 云判识开启,晴空点中 靶区/环境 2 均匀过滤后, 精匹配点个数 %d' % (
Band1, flag, len(idx[0]))
condition = np.logical_and(homoValue2 > 0, condition)
condition = np.logical_and(homoValue2 < th_vaue_max, condition)
idx = np.where(condition)
print u'%s %s 云判识开启,晴空点中 饱和值1 均匀过滤后, 精匹配点个数 %d' % (Band1, flag,
len(idx[0]))
idx_clear = np.where(condition)
self.MaskFine[Band1][idx_clear] = 1
# 云区判别
if 'ir' in flag:
condition = np.logical_and(self.MaskRough > 0,
irValue < modeCfg.clear_min_ir)
condition = np.logical_and(irValue > 0, condition)
elif 'vis' in flag:
condition = np.logical_and(
self.MaskRough > 0, visValue >= modeCfg.clear_max_vis)
idx = np.where(condition)
print u'%s %s 云判识开启,云区点个数 %d' % (Band1, flag, len(idx[0]))
condition = np.logical_and(SatZRaio < th_cld1, condition)
idx = np.where(condition)
print u'%s %s 云判识开启,云区点中 角度 均匀过滤后, 精匹配点个数 %d' % (Band1, flag,
len(idx[0]))
condition = np.logical_and(homoFov1 < th_cld2, condition)
idx = np.where(condition)
print u'%s %s 云判识开启,云区点中 靶区1 均匀过滤后, 精匹配点个数 %d' % (Band1, flag,
len(idx[0]))
condition = np.logical_and(homoEnv1 < th_cld3, condition)
idx = np.where(condition)
print u'%s %s 云判识开启,云区点中 环境1 均匀过滤后, 精匹配点个数 %d' % (Band1, flag,
len(idx[0]))
condition = np.logical_and(homoFovEnv1 < th_cld4, condition)
idx = np.where(condition)
print u'%s %s 云判识开启,云区点中 靶区/环境 1 均匀过滤后, 精匹配点个数 %d' % (
Band1, flag, len(idx[0]))
condition = np.logical_and(homoValue1 > 0, condition)
condition = np.logical_and(homoValue1 < th_vaue_max, condition)
idx = np.where(condition)
print u'%s %s 云判识开启,云区点中 饱和值1 均匀过滤后, 精匹配点个数 %d' % (Band1, flag,
len(idx[0]))
condition = np.logical_and(homoFov2 < th_cld2, condition)
idx = np.where(condition)
print u'%s %s 云判识开启,云区点中 靶区2 均匀过滤后, 精匹配点个数 %d' % (Band1, flag,
len(idx[0]))
condition = np.logical_and(homoEnv2 < th_cld3, condition)
idx = np.where(condition)
print u'%s %s 云判识开启,云区点中 环境2 均匀过滤后, 精匹配点个数 %d' % (Band1, flag,
len(idx[0]))
condition = np.logical_and(homoFovEnv2 < th_cld4, condition)
idx = np.where(condition)
print u'%s %s 云判识开启,云区点中 靶区/环境 2 均匀过滤后, 精匹配点个数 %d' % (
Band1, flag, len(idx[0]))
condition = np.logical_and(homoValue2 > 0, condition)
condition = np.logical_and(homoValue2 < th_vaue_max, condition)
idx = np.where(condition)
print u'%s %s 云判识开启,云区点中 饱和值1 均匀过滤后, 精匹配点个数 %d' % (Band1, flag,
len(idx[0]))
idx_cloud = np.where(condition)
totalNums = len(idx_cloud[0]) + len(idx_clear[0])
print u'%s %s 云判识开启，匹配点个数，晴空 %d 云区 %d 总计：%d' % (
Band1, flag, len(idx_clear[0]), len(
idx_cloud[0]), totalNums)
self.MaskFine[Band1][idx_cloud] = 1
def save_fine_data(self, modeCfg):
"""
第二轮匹配，根据各通道的的mean和std计以为，角度和距离等进行精细化过滤
"""
# 最终的公共匹配点数量
idx = np.where(self.PubIdx > 0)
if len(idx[0]) == 0:
return
print u'所有粗匹配点数目 ', len(idx[0])
# 掩码清零
self.MaskRough[:] = 0
# 计算共同区域的距离差 #########
disDiff = np.full_like(self.MERSI_Lons, '-1', dtype='i2')
a = np.power(self.MODIS_Lons[idx] - self.MERSI_Lons[idx], 2)
b = np.power(self.MODIS_Lats[idx] - self.MERSI_Lats[idx], 2)
disDiff[idx] = np.sqrt(a + b) * 100.
idx_Rough = np.logical_and(disDiff < modeCfg.distdif_max, disDiff >= 0)
idx1 = np.where(idx_Rough)
print u'1. 距离过滤后剩余点 ', len(idx1[0])
# 过滤太阳天顶角 ###############
idx_Rough = np.logical_and(
idx_Rough, self.MERSI_SoZ <= modeCfg.solzenith_max)
idx1 = np.where(idx_Rough)
print u'2. 太阳天顶角过滤后剩余点 ', len(idx1[0])
# 计算耀斑角 ###############
glint1 = np.full_like(self.MERSI_SatZ, -999.)
glint1[idx] = sun_glint_cal(
self.MERSI_SatA[idx], self.MERSI_SatZ[idx], self.MERSI_SoA[idx], self.MERSI_SoZ[idx])
idx_Rough = np.logical_and(idx_Rough, glint1 > modeCfg.solglint_min)
print np.nanmin(glint1[idx]), np.nanmax(glint1[idx])
idx1 = np.where(idx_Rough)
print u'3. 太阳耀斑角过滤后剩余点 ', len(idx1[0])
idx_Rough = np.logical_and(
idx_Rough, self.MERSI_SatZ <= modeCfg.satzenith_max)
idx1 = np.where(idx_Rough)
print u'4. FY卫星观测角(天顶角)滤后剩余点 ', len(idx1[0])
self.MaskRough[idx1] = 1
for Band1 in modeCfg.chan1:
# 掩码清零
self.MaskFine[Band1][:] = 0
th_vaue_max = modeCfg.CH_threshold[Band1]['value_max']
th2 = modeCfg.CH_threshold[Band1]['homodif_fov_max']
flag = 0
# 可见光通道
if hasattr(self, 'MERSI_FovMean') and Band1 in self.MERSI_FovMean.keys():
flag = 'vis'
print '1111', Band1
homoFov1 = np.abs(
self.MERSI_FovStd[Band1] / self.MERSI_FovMean[Band1])
homoValue1 = self.MERSI_FovMean[Band1]
homoFov2 = np.abs(
self.MODIS_FovStd[Band1] / self.MODIS_FovMean[Band1])
homoValue2 = self.MODIS_FovMean[Band1]
condition = np.logical_and(self.MaskRough > 0, True)
condition = np.logical_and(homoValue1 < th_vaue_max, condition)
condition = np.logical_and(homoValue1 > 0, condition)
idx = np.where(condition)
print u'%s %s 饱和值过滤后，精匹配点个数 %d' % (Band1, flag, len(idx[0]))
condition = np.logical_and(homoFov1 < th2, condition)
idx = np.where(condition)
print u'%s %s 靶区过滤后，精匹配点个数 %d' % (Band1, flag, len(idx[0]))
# sat 2过滤
condition = np.logical_and(homoValue2 > 0, condition)
condition = np.logical_and(homoValue2 < th_vaue_max, condition)
idx = np.where(condition)
print u'%s %s 饱和值2过滤后，精匹配点个数 %d' % (Band1, flag, len(idx[0]))
condition = np.logical_and(homoFov2 < th2, condition)
idx = np.where(condition)
print u'%s %s 靶区2过滤后，精匹配点个数 %d' % (Band1, flag, len(idx[0]))
self.MaskFine[Band1][idx] = 1