def generateCalibrationReport(config,
night_dir_path,
match_radius=2.0,
platepar=None,
show_graphs=False):
""" Given the folder of the night, find the Calstars file, check the star fit and generate a report
with the quality of the calibration. The report contains information about both the astrometry and
the photometry calibration. Graphs will be saved in the given directory of the night.
Arguments:
config: [Config instance]
night_dir_path: [str] Full path to the directory of the night.
Keyword arguments:
match_radius: [float] Match radius for star matching between image and catalog stars (px).
platepar: [Platepar instance] Use this platepar instead of finding one in the folder.
show_graphs: [bool] Show the graphs on the screen. False by default.
Return:
None
"""
# Find the CALSTARS file in the given folder
calstars_file = None
for calstars_file in os.listdir(night_dir_path):
if ('CALSTARS' in calstars_file) and ('.txt' in calstars_file):
break
if calstars_file is None:
print('CALSTARS file could not be found in the given directory!')
return None
# Load the calstars file
star_list = readCALSTARS(night_dir_path, calstars_file)
### Load recalibrated platepars, if they exist ###
# Find recalibrated platepars file per FF file
platepars_recalibrated_file = None
for file_name in os.listdir(night_dir_path):
if file_name == config.platepars_recalibrated_name:
platepars_recalibrated_file = file_name
break
# Load all recalibrated platepars if the file is available
recalibrated_platepars = None
if platepars_recalibrated_file:
with open(os.path.join(night_dir_path,
platepars_recalibrated_file)) as f:
recalibrated_platepars = json.load(f)
print(
'Loaded recalibrated platepars JSON file for the calibration report...'
)
### ###
### Load the platepar file ###
# Find the platepar file in the given directory if it was not given
if platepar is None:
# Find the platepar file
platepar_file = None
for file_name in os.listdir(night_dir_path):
if file_name == config.platepar_name:
platepar_file = file_name
break
if platepar_file is None:
print('The platepar cannot be found in the night directory!')
return None
# Load the platepar file
platepar = Platepar()
platepar.read(os.path.join(night_dir_path, platepar_file),
use_flat=config.use_flat)
### ###
night_name = os.path.split(night_dir_path.strip(os.sep))[1]
# Go one mag deeper than in the config
lim_mag = config.catalog_mag_limit + 1
# Load catalog stars (load one magnitude deeper)
catalog_stars, mag_band_str, config.star_catalog_band_ratios = StarCatalog.readStarCatalog(\
config.star_catalog_path, config.star_catalog_file, lim_mag=lim_mag, \
mag_band_ratios=config.star_catalog_band_ratios)
### Take only those CALSTARS entires for which FF files exist in the folder ###
# Get a list of FF files in the folder
ff_list = []
for file_name in os.listdir(night_dir_path):
if validFFName(file_name):
ff_list.append(file_name)
# Filter out calstars entries, generate a star dictionary where the keys are JDs of FFs
star_dict = {}
ff_dict = {}
for entry in star_list:
ff_name, star_data = entry
# Check if the FF from CALSTARS exists in the folder
if ff_name not in ff_list:
continue
dt = getMiddleTimeFF(ff_name, config.fps, ret_milliseconds=True)
jd = date2JD(*dt)
# Add the time and the stars to the dict
star_dict[jd] = star_data
ff_dict[jd] = ff_name
### ###
# If there are no FF files in the directory, don't generate a report
if len(star_dict) == 0:
print('No FF files from the CALSTARS file in the directory!')
return None
# If the recalibrated platepars file exists, take the one with the most stars
max_jd = 0
using_recalib_platepars = False
if recalibrated_platepars is not None:
max_stars = 0
for ff_name_temp in recalibrated_platepars:
# Compute the Julian date of the FF middle
dt = getMiddleTimeFF(ff_name_temp,
config.fps,
ret_milliseconds=True)
jd = date2JD(*dt)
# Check that this file exists in CALSTARS and the list of FF files
if (jd not in star_dict) or (jd not in ff_dict):
continue
# Check if the number of stars on this FF file is larger than the before
if len(star_dict[jd]) > max_stars:
max_jd = jd
max_stars = len(star_dict[jd])
# Set a flag to indicate if using recalibrated platepars has failed
if max_jd == 0:
using_recalib_platepars = False
else:
print('Using recalibrated platepars, file:', ff_dict[max_jd])
using_recalib_platepars = True
# Select the platepar where the FF file has the most stars
platepar_dict = recalibrated_platepars[ff_dict[max_jd]]
platepar = Platepar()
platepar.loadFromDict(platepar_dict, use_flat=config.use_flat)
filtered_star_dict = {max_jd: star_dict[max_jd]}
# Match stars on the image with the stars in the catalog
n_matched, avg_dist, cost, matched_stars = matchStarsResiduals(config, platepar, catalog_stars, \
filtered_star_dict, match_radius, ret_nmatch=True, lim_mag=lim_mag)
max_matched_stars = n_matched
# Otherwise take the optimal FF file for evaluation
if (recalibrated_platepars is None) or (not using_recalib_platepars):
# If there are more than a set number of FF files to evaluate, choose only the ones with most stars on
# the image
if len(star_dict) > config.calstars_files_N:
# Find JDs of FF files with most stars on them
top_nstars_indices = np.argsort([len(x) for x in star_dict.values()])[::-1][:config.calstars_files_N \
- 1]
filtered_star_dict = {}
for i in top_nstars_indices:
filtered_star_dict[list(star_dict.keys())[i]] = list(
star_dict.values())[i]
star_dict = filtered_star_dict
# Match stars on the image with the stars in the catalog
n_matched, avg_dist, cost, matched_stars = matchStarsResiduals(config, platepar, catalog_stars, \
star_dict, match_radius, ret_nmatch=True, lim_mag=lim_mag)
# If no recalibrated platepars where found, find the image with the largest number of matched stars
if (not using_recalib_platepars) or (max_jd == 0):
max_jd = 0
max_matched_stars = 0
for jd in matched_stars:
_, _, distances = matched_stars[jd]
if len(distances) > max_matched_stars:
max_jd = jd
max_matched_stars = len(distances)
# If there are no matched stars, use the image with the largest number of detected stars
if max_matched_stars <= 2:
max_jd = max(star_dict, key=lambda x: len(star_dict[x]))
distances = [np.inf]
# Take the FF file with the largest number of matched stars
ff_name = ff_dict[max_jd]
# Load the FF file
ff = readFF(night_dir_path, ff_name)
img_h, img_w = ff.avepixel.shape
dpi = 200
plt.figure(figsize=(ff.avepixel.shape[1] / dpi,
ff.avepixel.shape[0] / dpi),
dpi=dpi)
# Take the average pixel
img = ff.avepixel
# Slightly adjust the levels
img = Image.adjustLevels(img, np.percentile(img, 1.0), 1.3,
np.percentile(img, 99.99))
plt.imshow(img, cmap='gray', interpolation='nearest')
legend_handles = []
# Plot detected stars
for img_star in star_dict[max_jd]:
y, x, _, _ = img_star
rect_side = 5 * match_radius
square_patch = plt.Rectangle((x - rect_side/2, y - rect_side/2), rect_side, rect_side, color='g', \
fill=False, label='Image stars')
plt.gca().add_artist(square_patch)
legend_handles.append(square_patch)
# If there are matched stars, plot them
if max_matched_stars > 2:
# Take the solution with the largest number of matched stars
image_stars, matched_catalog_stars, distances = matched_stars[max_jd]
# Plot matched stars
for img_star in image_stars:
x, y, _, _ = img_star
circle_patch = plt.Circle((y, x), radius=3*match_radius, color='y', fill=False, \
label='Matched stars')
plt.gca().add_artist(circle_patch)
legend_handles.append(circle_patch)
### Plot match residuals ###
# Compute preducted positions of matched image stars from the catalog
x_predicted, y_predicted = raDecToXYPP(matched_catalog_stars[:, 0], \
matched_catalog_stars[:, 1], max_jd, platepar)
img_y, img_x, _, _ = image_stars.T
delta_x = x_predicted - img_x
delta_y = y_predicted - img_y
# Compute image residual and angle of the error
res_angle = np.arctan2(delta_y, delta_x)
res_distance = np.sqrt(delta_x**2 + delta_y**2)
# Calculate coordinates of the beginning of the residual line
res_x_beg = img_x + 3 * match_radius * np.cos(res_angle)
res_y_beg = img_y + 3 * match_radius * np.sin(res_angle)
# Calculate coordinates of the end of the residual line
res_x_end = img_x + 100 * np.cos(res_angle) * res_distance
res_y_end = img_y + 100 * np.sin(res_angle) * res_distance
# Plot the 100x residuals
for i in range(len(x_predicted)):
res_plot = plt.plot([res_x_beg[i], res_x_end[i]], [res_y_beg[i], res_y_end[i]], color='orange', \
lw=0.5, label='100x residuals')
legend_handles.append(res_plot[0])
### ###
else:
distances = [np.inf]
# If there are no matched stars, plot large text in the middle of the screen
plt.text(img_w / 2,
img_h / 2,
"NO MATCHED STARS!",
color='r',
alpha=0.5,
fontsize=20,
ha='center',
va='center')
### Plot positions of catalog stars to the limiting magnitude of the faintest matched star + 1 mag ###
# Find the faintest magnitude among matched stars
if max_matched_stars > 2:
faintest_mag = np.max(matched_catalog_stars[:, 2]) + 1
else:
# If there are no matched stars, use the limiting magnitude from config
faintest_mag = config.catalog_mag_limit + 1
# Estimate RA,dec of the centre of the FOV
_, RA_c, dec_c, _ = xyToRaDecPP([jd2Date(max_jd)], [platepar.X_res / 2],
[platepar.Y_res / 2], [1], platepar)
RA_c = RA_c[0]
dec_c = dec_c[0]
fov_radius = np.hypot(*computeFOVSize(platepar))
# Get stars from the catalog around the defined center in a given radius
_, extracted_catalog = subsetCatalog(catalog_stars, RA_c, dec_c,
fov_radius, faintest_mag)
ra_catalog, dec_catalog, mag_catalog = extracted_catalog.T
# Compute image positions of all catalog stars that should be on the image
x_catalog, y_catalog = raDecToXYPP(ra_catalog, dec_catalog, max_jd,
platepar)
# Filter all catalog stars outside the image
temp_arr = np.c_[x_catalog, y_catalog, mag_catalog]
temp_arr = temp_arr[temp_arr[:, 0] >= 0]
temp_arr = temp_arr[temp_arr[:, 0] <= ff.avepixel.shape[1]]
temp_arr = temp_arr[temp_arr[:, 1] >= 0]
temp_arr = temp_arr[temp_arr[:, 1] <= ff.avepixel.shape[0]]
x_catalog, y_catalog, mag_catalog = temp_arr.T
# Plot catalog stars on the image
cat_stars_handle = plt.scatter(x_catalog, y_catalog, c='none', marker='D', lw=1.0, alpha=0.4, \
s=((4.0 + (faintest_mag - mag_catalog))/3.0)**(2*2.512), edgecolor='r', label='Catalog stars')
legend_handles.append(cat_stars_handle)
### ###
# Add info text in the corner
info_text = ff_dict[max_jd] + '\n' \
+ "Matched stars within {:.1f} px radius: {:d}/{:d} \n".format(match_radius, max_matched_stars, \
len(star_dict[max_jd])) \
+ "Median distance = {:.2f} px\n".format(np.median(distances)) \
+ "Catalog lim mag = {:.1f}".format(lim_mag)
plt.text(10, 10, info_text, bbox=dict(facecolor='black', alpha=0.5), va='top', ha='left', fontsize=4, \
color='w', family='monospace')
legend = plt.legend(handles=legend_handles,
prop={'size': 4},
loc='upper right')
legend.get_frame().set_facecolor('k')
legend.get_frame().set_edgecolor('k')
for txt in legend.get_texts():
txt.set_color('w')
### Add FOV info (centre, size) ###
# Mark FOV centre
plt.scatter(platepar.X_res / 2,
platepar.Y_res / 2,
marker='+',
s=20,
c='r',
zorder=4)
# Compute FOV centre alt/az
azim_centre, alt_centre = raDec2AltAz(max_jd, platepar.lon, platepar.lat,
RA_c, dec_c)
# Compute FOV size
fov_h, fov_v = computeFOVSize(platepar)
# Compute the rotation wrt. horizon
rot_horizon = rotationWrtHorizon(platepar)
fov_centre_text = "Azim = {:6.2f}$\\degree$\n".format(azim_centre) \
+ "Alt = {:6.2f}$\\degree$\n".format(alt_centre) \
+ "Rot h = {:6.2f}$\\degree$\n".format(rot_horizon) \
+ "FOV h = {:6.2f}$\\degree$\n".format(fov_h) \
+ "FOV v = {:6.2f}$\\degree$".format(fov_v) \
plt.text(10, platepar.Y_res - 10, fov_centre_text, bbox=dict(facecolor='black', alpha=0.5), \
va='bottom', ha='left', fontsize=4, color='w', family='monospace')
### ###
# Plot RA/Dec gridlines #
addEquatorialGrid(plt, platepar, max_jd)
plt.axis('off')
plt.gca().get_xaxis().set_visible(False)
plt.gca().get_yaxis().set_visible(False)
plt.xlim([0, ff.avepixel.shape[1]])
plt.ylim([ff.avepixel.shape[0], 0])
# Remove the margins
plt.subplots_adjust(left=0, bottom=0, right=1, top=1, wspace=0, hspace=0)
plt.savefig(os.path.join(night_dir_path, night_name + '_calib_report_astrometry.jpg'), \
bbox_inches='tight', pad_inches=0, dpi=dpi)
if show_graphs:
plt.show()
else:
plt.clf()
plt.close()
if max_matched_stars > 2:
### PHOTOMETRY FIT ###
# If a flat is used, set the vignetting coeff to 0
if config.use_flat:
platepar.vignetting_coeff = 0.0
# Extact intensities and mangitudes
star_intensities = image_stars[:, 2]
catalog_mags = matched_catalog_stars[:, 2]
# Compute radius of every star from image centre
radius_arr = np.hypot(image_stars[:, 0] - img_h / 2,
image_stars[:, 1] - img_w / 2)
# Fit the photometry on automated star intensities (use the fixed vignetting coeff, use robust fit)
photom_params, fit_stddev, fit_resid, star_intensities, radius_arr, catalog_mags = \
photometryFitRobust(star_intensities, radius_arr, catalog_mags, \
fixed_vignetting=platepar.vignetting_coeff)
photom_offset, _ = photom_params
### ###
### PLOT PHOTOMETRY ###
# Note: An almost identical code exists in RMS.Astrometry.SkyFit in the PlateTool.photometry function
dpi = 130
fig_p, (ax_p, ax_r) = plt.subplots(nrows=2, facecolor=None, figsize=(6.0, 7.0), dpi=dpi, \
gridspec_kw={'height_ratios':[2, 1]})
# Plot raw star intensities
ax_p.scatter(-2.5 * np.log10(star_intensities),
catalog_mags,
s=5,
c='r',
alpha=0.5,
label="Raw")
# If a flat is used, disregard the vignetting
if not config.use_flat:
# Plot intensities of image stars corrected for vignetting
lsp_corr_arr = np.log10(correctVignetting(star_intensities, radius_arr, \
platepar.vignetting_coeff))
ax_p.scatter(-2.5*lsp_corr_arr, catalog_mags, s=5, c='b', alpha=0.5, \
label="Corrected for vignetting")
# Plot photometric offset from the platepar
x_min, x_max = ax_p.get_xlim()
y_min, y_max = ax_p.get_ylim()
x_min_w = x_min - 3
x_max_w = x_max + 3
y_min_w = y_min - 3
y_max_w = y_max + 3
photometry_info = "Platepar: {:+.1f}*LSP + {:.2f} +/- {:.2f}".format(platepar.mag_0, \
platepar.mag_lev, platepar.mag_lev_stddev) \
+ "\nVignetting coeff = {:.5f}".format(platepar.vignetting_coeff) \
+ "\nGamma = {:.2f}".format(platepar.gamma)
# Plot the photometry calibration from the platepar
logsum_arr = np.linspace(x_min_w, x_max_w, 10)
ax_p.plot(logsum_arr, logsum_arr + platepar.mag_lev, label=photometry_info, linestyle='--', \
color='k', alpha=0.5)
# Plot the fitted photometry calibration
fit_info = "Fit: {:+.1f}*LSP + {:.2f} +/- {:.2f}".format(
-2.5, photom_offset, fit_stddev)
ax_p.plot(logsum_arr,
logsum_arr + photom_offset,
label=fit_info,
linestyle='--',
color='b',
alpha=0.75)
ax_p.legend()
ax_p.set_ylabel("Catalog magnitude ({:s})".format(mag_band_str))
ax_p.set_xlabel("Uncalibrated magnitude")
# Set wider axis limits
ax_p.set_xlim(x_min_w, x_max_w)
ax_p.set_ylim(y_min_w, y_max_w)
ax_p.invert_yaxis()
ax_p.invert_xaxis()
ax_p.grid()
### Plot photometry vs radius ###
img_diagonal = np.hypot(img_h / 2, img_w / 2)
# Plot photometry residuals (including vignetting)
ax_r.scatter(radius_arr, fit_resid, c='b', alpha=0.75, s=5, zorder=3)
# Plot a zero line
ax_r.plot(np.linspace(0, img_diagonal, 10), np.zeros(10), linestyle='dashed', alpha=0.5, \
color='k')
# Plot only when no flat is used
if not config.use_flat:
# Plot radius from centre vs. fit residual
fit_resids_novignetting = catalog_mags - photomLine((np.array(star_intensities), \
np.array(radius_arr)), photom_offset, 0.0)
ax_r.scatter(radius_arr,
fit_resids_novignetting,
s=5,
c='r',
alpha=0.5,
zorder=3)
px_sum_tmp = 1000
radius_arr_tmp = np.linspace(0, img_diagonal, 50)
# Plot vignetting loss curve
vignetting_loss = 2.5*np.log10(px_sum_tmp) \
- 2.5*np.log10(correctVignetting(px_sum_tmp, radius_arr_tmp, \
platepar.vignetting_coeff))
ax_r.plot(radius_arr_tmp,
vignetting_loss,
linestyle='dotted',
alpha=0.5,
color='k')
ax_r.grid()
ax_r.set_ylabel("Fit residuals (mag)")
ax_r.set_xlabel("Radius from centre (px)")
ax_r.set_xlim(0, img_diagonal)
### ###
plt.tight_layout()
plt.savefig(os.path.join(night_dir_path,
night_name + '_calib_report_photometry.png'),
dpi=150)
if show_graphs:
plt.show()
else:
plt.clf()
plt.close()
def addEquatorialGrid(plt_handle, platepar, jd):
""" Given the plot handle containing the image, the function plots an equatorial grid.
Arguments:
plt_handle: [pyplot instance]
platepar: [Platepar object]
jd: [float] Julian date of the image.
Return:
plt_handle: [pyplot instance] Pyplot instance with the added grid.
"""
# Estimate RA,dec of the centre of the FOV
_, RA_c, dec_c, _ = xyToRaDecPP([jd2Date(jd)], [platepar.X_res / 2],
[platepar.Y_res / 2], [1],
platepar,
extinction_correction=False)
RA_c = RA_c[0]
dec_c = dec_c[0]
# Compute FOV centre alt/az
azim_centre, alt_centre = raDec2AltAz(RA_c, dec_c, jd, platepar.lat,
platepar.lon)
# Compute FOV size
fov_h, fov_v = computeFOVSize(platepar)
fov_radius = np.hypot(*computeFOVSize(platepar))
# Determine gridline frequency (double the gridlines if the number is < 4eN)
grid_freq = 10**np.floor(np.log10(fov_radius))
if 10**(np.log10(fov_radius) - np.floor(np.log10(fov_radius))) < 4:
grid_freq /= 2
# Set a maximum grid frequency of 15 deg
if grid_freq > 15:
grid_freq = 15
# Grid plot density
plot_dens = grid_freq / 100
# Compute the range of declinations to consider
dec_min = platepar.dec_d - fov_radius / 2
if dec_min < -90:
dec_min = -90
dec_max = platepar.dec_d + fov_radius / 2
if dec_max > 90:
dec_max = 90
ra_grid_arr = np.arange(0, 360, grid_freq)
dec_grid_arr = np.arange(-90, 90, grid_freq)
# Filter out the dec grid for min/max declination
dec_grid_arr = dec_grid_arr[(dec_grid_arr >= dec_min)
& (dec_grid_arr <= dec_max)]
# Plot the celestial parallel grid
for dec_grid in dec_grid_arr:
ra_grid_plot = np.arange(0, 360, plot_dens)
dec_grid_plot = np.zeros_like(ra_grid_plot) + dec_grid
# Compute alt/az
az_grid_plot, alt_grid_plot = raDec2AltAz_vect(ra_grid_plot, dec_grid_plot, jd, platepar.lat, \
platepar.lon)
# Filter out points below the horizon and outside the FOV
filter_arr = (alt_grid_plot > 0) & (np.degrees(angularSeparation(np.radians(alt_centre), \
np.radians(azim_centre), np.radians(alt_grid_plot), np.radians(az_grid_plot))) < fov_radius)
ra_grid_plot = ra_grid_plot[filter_arr]
dec_grid_plot = dec_grid_plot[filter_arr]
# Find gaps in continuity and break up plotting individual lines
gap_indices = np.argwhere(
np.abs(ra_grid_plot[1:] - ra_grid_plot[:-1]) > fov_radius)
if len(gap_indices):
ra_grid_plot_list = []
dec_grid_plot_list = []
# Separate gridlines with large gaps
prev_gap_indx = 0
for entry in gap_indices:
gap_indx = entry[0]
ra_grid_plot_list.append(ra_grid_plot[prev_gap_indx:gap_indx +
1])
dec_grid_plot_list.append(
dec_grid_plot[prev_gap_indx:gap_indx + 1])
prev_gap_indx = gap_indx
# Add the last segment
ra_grid_plot_list.append(ra_grid_plot[prev_gap_indx + 1:-1])
dec_grid_plot_list.append(dec_grid_plot[prev_gap_indx + 1:-1])
else:
ra_grid_plot_list = [ra_grid_plot]
dec_grid_plot_list = [dec_grid_plot]
# Plot all grid segments
for ra_grid_plot, dec_grid_plot in zip(ra_grid_plot_list,
dec_grid_plot_list):
# Compute image coordinates for every grid celestial parallel
x_grid, y_grid = raDecToXYPP(ra_grid_plot, dec_grid_plot, jd,
platepar)
# Plot the grid
plt_handle.plot(x_grid,
y_grid,
color='w',
alpha=0.2,
zorder=2,
linewidth=0.5,
linestyle='dotted')
# Plot the celestial meridian grid
for ra_grid in ra_grid_arr:
dec_grid_plot = np.arange(-90, 90, plot_dens)
ra_grid_plot = np.zeros_like(dec_grid_plot) + ra_grid
# Filter out the dec grid
filter_arr = (dec_grid_plot >= dec_min) & (dec_grid_plot <= dec_max)
ra_grid_plot = ra_grid_plot[filter_arr]
dec_grid_plot = dec_grid_plot[filter_arr]
# Compute alt/az
az_grid_plot, alt_grid_plot = raDec2AltAz_vect(ra_grid_plot, dec_grid_plot, jd, platepar.lat, \
platepar.lon)
# Filter out points below the horizon
filter_arr = (alt_grid_plot > 0) & (np.degrees(angularSeparation(np.radians(alt_centre), \
np.radians(azim_centre), np.radians(alt_grid_plot), np.radians(az_grid_plot))) < fov_radius)
ra_grid_plot = ra_grid_plot[filter_arr]
dec_grid_plot = dec_grid_plot[filter_arr]
# Compute image coordinates for every grid celestial parallel
x_grid, y_grid = raDecToXYPP(ra_grid_plot, dec_grid_plot, jd, platepar)
# # Filter out everything outside the FOV
# filter_arr = (x_grid >= 0) & (x_grid <= platepar.X_res) & (y_grid >= 0) & (y_grid <= platepar.Y_res)
# x_grid = x_grid[filter_arr]
# y_grid = y_grid[filter_arr]
# Plot the grid
plt_handle.plot(x_grid,
y_grid,
color='w',
alpha=0.2,
zorder=2,
linewidth=0.5,
linestyle='dotted')
return plt_handle
def updateRaDecGrid(grid, platepar):
"""
Updates the values of grid to form a right ascension and declination grid
Arguments:
grid: [pg.PlotCurveItem]
platepar: [Platepar object]
"""
# Estimate RA,dec of the centre of the FOV
_, RA_c, dec_c, _ = xyToRaDecPP([jd2Date(platepar.JD)],
[platepar.X_res / 2], [platepar.Y_res / 2],
[1],
platepar,
extinction_correction=False)
# azim_centre, alt_centre = trueRaDec2ApparentAltAz(RA_c, dec_c, platepar.JD, platepar.lat, platepar.lon)
# Compute FOV size
fov_radius = np.hypot(*computeFOVSize(platepar))
# Determine gridline frequency (double the gridlines if the number is < 4eN)
grid_freq = 10**np.floor(np.log10(fov_radius))
if 10**(np.log10(fov_radius) - np.floor(np.log10(fov_radius))) < 4:
grid_freq /= 2
# Set a maximum grid frequency of 15 deg
if grid_freq > 15:
grid_freq = 15
# Grid plot density
plot_dens = grid_freq / 100
ra_grid_arr = np.arange(0, 360, grid_freq)
dec_grid_arr = np.arange(-90, 90, grid_freq)
x = []
y = []
cuts = []
# Plot the celestial parallel grid (circles)
for dec_grid in dec_grid_arr:
ra_grid_plot = np.arange(0, 360, plot_dens)
dec_grid_plot = np.zeros_like(ra_grid_plot) + dec_grid
# Compute alt/az
az_grid_plot, alt_grid_plot = trueRaDec2ApparentAltAz(
ra_grid_plot, dec_grid_plot, platepar.JD, platepar.lat,
platepar.lon, platepar.refraction)
# Filter out points below the horizon and outside the FOV
filter_arr = (alt_grid_plot > 0) # & (angularSeparation(alt_centre,
# azim_centre,
# alt_grid_plot,
# az_grid_plot) < fov_radius)
ra_grid_plot = ra_grid_plot[filter_arr]
dec_grid_plot = dec_grid_plot[filter_arr]
# Compute image coordinates for every grid celestial parallel
x_grid, y_grid = raDecToXYPP(ra_grid_plot, dec_grid_plot, platepar.JD,
platepar)
filter_arr = (x_grid >= 0) & (x_grid <= platepar.X_res) & (
y_grid >= 0) & (y_grid <= platepar.Y_res)
x_grid = x_grid[filter_arr]
y_grid = y_grid[filter_arr]
x.extend(x_grid)
y.extend(y_grid)
cuts.append(len(x) - 1)
# Plot the celestial meridian grid (outward lines)
for ra_grid in ra_grid_arr:
dec_grid_plot = np.arange(-90, 90, plot_dens) # how close to horizon
ra_grid_plot = np.zeros_like(dec_grid_plot) + ra_grid
# Compute alt/az
az_grid_plot, alt_grid_plot = trueRaDec2ApparentAltAz(
ra_grid_plot, dec_grid_plot, platepar.JD, platepar.lat,
platepar.lon, platepar.refraction)
# Filter out points below the horizon
filter_arr = (alt_grid_plot > 0) #& (angularSeparation(alt_centre,
# azim_centre,
# alt_grid_plot,
# az_grid_plot) < fov_radius)
ra_grid_plot = ra_grid_plot[filter_arr]
dec_grid_plot = dec_grid_plot[filter_arr]
# Compute image coordinates for every grid celestial parallel
x_grid, y_grid = raDecToXYPP(ra_grid_plot, dec_grid_plot, platepar.JD,
platepar)
filter_arr = (x_grid >= 0) & (x_grid <= platepar.X_res) & (
y_grid >= 0) & (y_grid <= platepar.Y_res)
x_grid = x_grid[filter_arr]
y_grid = y_grid[filter_arr]
x.extend(x_grid)
y.extend(y_grid)
cuts.append(len(x) - 1)
# horizon
az_horiz_arr = np.arange(0, 360, plot_dens)
alt_horiz_arr = np.zeros_like(az_horiz_arr)
ra_horiz_plot, dec_horiz_plot = apparentAltAz2TrueRADec(
az_horiz_arr, alt_horiz_arr, platepar.JD, platepar.lat, platepar.lon,
platepar.refraction)
x_horiz, y_horiz = raDecToXYPP(ra_horiz_plot, dec_horiz_plot, platepar.JD,
platepar)
filter_arr = (x_horiz >= 0) & (x_horiz <= platepar.X_res) & (
y_horiz >= 0) & (y_horiz <= platepar.Y_res)
x_horiz = x_horiz[filter_arr]
y_horiz = y_horiz[filter_arr]
x.extend(x_horiz)
y.extend(y_horiz)
cuts.append(len(x) - 1)
r = 15 # adjust this parameter if you see extraneous lines
# disconnect lines that are distant (unfinished circles had straight lines completing them)
for i in range(len(x) - 1):
if (x[i] - x[i + 1])**2 + (y[i] - y[i + 1])**2 > r**2:
cuts.append(i)
# convert cuts into connect
connect = np.full(len(x), 1)
if len(connect) > 0:
for i in cuts:
connect[i] = 0
grid.setData(x=x, y=y, connect=connect)
def updateAzAltGrid(grid, platepar):
"""
Updates the values of grid to form an azimuth and altitude grid
Arguments:
grid: [pg.PlotCurveItem]
platepar: [Platepar object]
"""
# Compute FOV size
fov_radius = np.hypot(*computeFOVSize(platepar))
# Determine gridline frequency (double the gridlines if the number is < 4eN)
grid_freq = 10**np.floor(np.log10(fov_radius))
if 10**(np.log10(fov_radius) - np.floor(np.log10(fov_radius))) < 4:
grid_freq /= 2
# Set a maximum grid frequency of 15 deg
if grid_freq > 15:
grid_freq = 15
# Grid plot density
plot_dens = grid_freq / 100
az_grid_arr = np.arange(0, 90, grid_freq)
alt_grid_arr = np.arange(0, 360, grid_freq)
x = []
y = []
cuts = []
# circles
for az_grid in az_grid_arr:
alt_grid_plot = np.arange(0, 360,
plot_dens) # how many degrees of circle
az_grid_plot = np.zeros_like(alt_grid_plot) + az_grid
# filter_arr = (angularSeparation(alt_centre,
# azim_centre,
# alt_grid_plot,
# az_grid_plot) < fov_radius)
#
# alt_grid_plot = alt_grid_plot[filter_arr]
# az_grid_plot = az_grid_plot[filter_arr]
ra_grid_plot, dec_grid_plot = apparentAltAz2TrueRADec(
alt_grid_plot, az_grid_plot, platepar.JD, platepar.lat,
platepar.lon, platepar.refraction)
x_grid, y_grid = raDecToXYPP(ra_grid_plot, dec_grid_plot, platepar.JD,
platepar)
filter_arr = (x_grid >= 0) & (x_grid <= platepar.X_res) & (
y_grid >= 0) & (y_grid <= platepar.Y_res)
x_grid = x_grid[filter_arr]
y_grid = y_grid[filter_arr]
x.extend(x_grid)
y.extend(y_grid)
cuts.append(len(x) - 1)
# Outward lines
for alt_grid in alt_grid_arr:
az_grid_plot = np.arange(0, 90, plot_dens)
alt_grid_plot = np.zeros_like(az_grid_plot) + alt_grid
# filter_arr = (angularSeparation(alt_centre,
# azim_centre,
# alt_grid_plot,
# az_grid_plot) < fov_radius)
#
# alt_grid_plot = alt_grid_plot[filter_arr]
# az_grid_plot = az_grid_plot[filter_arr]
ra_grid_plot, dec_grid_plot = apparentAltAz2TrueRADec(
alt_grid_plot, az_grid_plot, platepar.JD, platepar.lat,
platepar.lon, platepar.refraction)
x_grid, y_grid = raDecToXYPP(ra_grid_plot, dec_grid_plot, platepar.JD,
platepar)
filter_arr = (x_grid >= 0) & (x_grid <= platepar.X_res) & (
y_grid >= 0) & (y_grid <= platepar.Y_res)
x_grid = x_grid[filter_arr]
y_grid = y_grid[filter_arr]
x.extend(x_grid)
y.extend(y_grid)
cuts.append(len(x) - 1)
r = 50
for i in range(len(x) - 1):
if (x[i] - x[i + 1])**2 + (y[i] - y[i + 1])**2 > r**2:
cuts.append(i)
connect = np.full(len(x), 1)
for i in cuts[:-1]:
connect[i] = 0
grid.setData(x=x, y=y, connect=connect)
def generateCalibrationReport(config, night_dir_path, match_radius=2.0, platepar=None, show_graphs=False):
""" Given the folder of the night, find the Calstars file, check the star fit and generate a report
with the quality of the calibration. The report contains information about both the astrometry and
the photometry calibration. Graphs will be saved in the given directory of the night.
Arguments:
config: [Config instance]
night_dir_path: [str] Full path to the directory of the night.
Keyword arguments:
match_radius: [float] Match radius for star matching between image and catalog stars (px).
platepar: [Platepar instance] Use this platepar instead of finding one in the folder.
show_graphs: [bool] Show the graphs on the screen. False by default.
Return:
None
"""
# Find the CALSTARS file in the given folder
calstars_file = None
for calstars_file in os.listdir(night_dir_path):
if ('CALSTARS' in calstars_file) and ('.txt' in calstars_file):
break
if calstars_file is None:
print('CALSTARS file could not be found in the given directory!')
return None
# Load the calstars file
star_list = readCALSTARS(night_dir_path, calstars_file)
### Load recalibrated platepars, if they exist ###
# Find recalibrated platepars file per FF file
platepars_recalibrated_file = None
for file_name in os.listdir(night_dir_path):
if file_name == config.platepars_recalibrated_name:
platepars_recalibrated_file = file_name
break
# Load all recalibrated platepars if the file is available
recalibrated_platepars = None
if platepars_recalibrated_file:
with open(os.path.join(night_dir_path, platepars_recalibrated_file)) as f:
recalibrated_platepars = json.load(f)
print('Loaded recalibrated platepars JSON file for the calibration report...')
### ###
### Load the platepar file ###
# Find the platepar file in the given directory if it was not given
if platepar is None:
# Find the platepar file
platepar_file = None
for file_name in os.listdir(night_dir_path):
if file_name == config.platepar_name:
platepar_file = file_name
break
if platepar_file is None:
print('The platepar cannot be found in the night directory!')
return None
# Load the platepar file
platepar = Platepar()
platepar.read(os.path.join(night_dir_path, platepar_file))
### ###
night_name = os.path.split(night_dir_path.strip(os.sep))[1]
# Go one mag deeper than in the config
lim_mag = config.catalog_mag_limit + 1
# Load catalog stars (load one magnitude deeper)
catalog_stars, mag_band_str, config.star_catalog_band_ratios = StarCatalog.readStarCatalog(\
config.star_catalog_path, config.star_catalog_file, lim_mag=lim_mag, \
mag_band_ratios=config.star_catalog_band_ratios)
### Take only those CALSTARS entires for which FF files exist in the folder ###
# Get a list of FF files in the folder\
ff_list = []
for file_name in os.listdir(night_dir_path):
if validFFName(file_name):
ff_list.append(file_name)
# Filter out calstars entries, generate a star dictionary where the keys are JDs of FFs
star_dict = {}
ff_dict = {}
for entry in star_list:
ff_name, star_data = entry
# Check if the FF from CALSTARS exists in the folder
if ff_name not in ff_list:
continue
dt = getMiddleTimeFF(ff_name, config.fps, ret_milliseconds=True)
jd = date2JD(*dt)
# Add the time and the stars to the dict
star_dict[jd] = star_data
ff_dict[jd] = ff_name
### ###
# If there are no FF files in the directory, don't generate a report
if len(star_dict) == 0:
print('No FF files from the CALSTARS file in the directory!')
return None
# If the recalibrated platepars file exists, take the one with the most stars
max_jd = 0
if recalibrated_platepars is not None:
max_stars = 0
for ff_name_temp in recalibrated_platepars:
# Compute the Julian date of the FF middle
dt = getMiddleTimeFF(ff_name_temp, config.fps, ret_milliseconds=True)
jd = date2JD(*dt)
# Check that this file exists in CALSTARS and the list of FF files
if (jd not in star_dict) or (jd not in ff_dict):
continue
# Check if the number of stars on this FF file is larger than the before
if len(star_dict[jd]) > max_stars:
max_jd = jd
max_stars = len(star_dict[jd])
# Set a flag to indicate if using recalibrated platepars has failed
if max_jd == 0:
using_recalib_platepars = False
else:
print('Using recalibrated platepars, file:', ff_dict[max_jd])
using_recalib_platepars = True
# Select the platepar where the FF file has the most stars
platepar_dict = recalibrated_platepars[ff_dict[max_jd]]
platepar = Platepar()
platepar.loadFromDict(platepar_dict)
filtered_star_dict = {max_jd: star_dict[max_jd]}
# Match stars on the image with the stars in the catalog
n_matched, avg_dist, cost, matched_stars = matchStarsResiduals(config, platepar, catalog_stars, \
filtered_star_dict, match_radius, ret_nmatch=True, lim_mag=lim_mag)
max_matched_stars = n_matched
# Otherwise take the optimal FF file for evaluation
if (recalibrated_platepars is None) or (not using_recalib_platepars):
# If there are more than a set number of FF files to evaluate, choose only the ones with most stars on
# the image
if len(star_dict) > config.calstars_files_N:
# Find JDs of FF files with most stars on them
top_nstars_indices = np.argsort([len(x) for x in star_dict.values()])[::-1][:config.calstars_files_N \
- 1]
filtered_star_dict = {}
for i in top_nstars_indices:
filtered_star_dict[list(star_dict.keys())[i]] = list(star_dict.values())[i]
star_dict = filtered_star_dict
# Match stars on the image with the stars in the catalog
n_matched, avg_dist, cost, matched_stars = matchStarsResiduals(config, platepar, catalog_stars, \
star_dict, match_radius, ret_nmatch=True, lim_mag=lim_mag)
# If no recalibrated platepars where found, find the image with the largest number of matched stars
if (not using_recalib_platepars) or (max_jd == 0):
max_jd = 0
max_matched_stars = 0
for jd in matched_stars:
_, _, distances = matched_stars[jd]
if len(distances) > max_matched_stars:
max_jd = jd
max_matched_stars = len(distances)
# If there are no matched stars, use the image with the largest number of detected stars
if max_matched_stars <= 2:
max_jd = max(star_dict, key=lambda x: len(star_dict[x]))
distances = [np.inf]
# Take the FF file with the largest number of matched stars
ff_name = ff_dict[max_jd]
# Load the FF file
ff = readFF(night_dir_path, ff_name)
img_h, img_w = ff.avepixel.shape
dpi = 200
plt.figure(figsize=(ff.avepixel.shape[1]/dpi, ff.avepixel.shape[0]/dpi), dpi=dpi)
# Take the average pixel
img = ff.avepixel
# Slightly adjust the levels
img = Image.adjustLevels(img, np.percentile(img, 1.0), 1.2, np.percentile(img, 99.99))
plt.imshow(img, cmap='gray', interpolation='nearest')
legend_handles = []
# Plot detected stars
for img_star in star_dict[max_jd]:
y, x, _, _ = img_star
rect_side = 5*match_radius
square_patch = plt.Rectangle((x - rect_side/2, y - rect_side/2), rect_side, rect_side, color='g', \
fill=False, label='Image stars')
plt.gca().add_artist(square_patch)
legend_handles.append(square_patch)
# If there are matched stars, plot them
if max_matched_stars > 2:
# Take the solution with the largest number of matched stars
image_stars, matched_catalog_stars, distances = matched_stars[max_jd]
# Plot matched stars
for img_star in image_stars:
x, y, _, _ = img_star
circle_patch = plt.Circle((y, x), radius=3*match_radius, color='y', fill=False, \
label='Matched stars')
plt.gca().add_artist(circle_patch)
legend_handles.append(circle_patch)
### Plot match residuals ###
# Compute preducted positions of matched image stars from the catalog
x_predicted, y_predicted = raDecToXYPP(matched_catalog_stars[:, 0], \
matched_catalog_stars[:, 1], max_jd, platepar)
img_y, img_x, _, _ = image_stars.T
delta_x = x_predicted - img_x
delta_y = y_predicted - img_y
# Compute image residual and angle of the error
res_angle = np.arctan2(delta_y, delta_x)
res_distance = np.sqrt(delta_x**2 + delta_y**2)
# Calculate coordinates of the beginning of the residual line
res_x_beg = img_x + 3*match_radius*np.cos(res_angle)
res_y_beg = img_y + 3*match_radius*np.sin(res_angle)
# Calculate coordinates of the end of the residual line
res_x_end = img_x + 100*np.cos(res_angle)*res_distance
res_y_end = img_y + 100*np.sin(res_angle)*res_distance
# Plot the 100x residuals
for i in range(len(x_predicted)):
res_plot = plt.plot([res_x_beg[i], res_x_end[i]], [res_y_beg[i], res_y_end[i]], color='orange', \
lw=0.5, label='100x residuals')
legend_handles.append(res_plot[0])
### ###
else:
distances = [np.inf]
# If there are no matched stars, plot large text in the middle of the screen
plt.text(img_w/2, img_h/2, "NO MATCHED STARS!", color='r', alpha=0.5, fontsize=20, ha='center',
va='center')
### Plot positions of catalog stars to the limiting magnitude of the faintest matched star + 1 mag ###
# Find the faintest magnitude among matched stars
if max_matched_stars > 2:
faintest_mag = np.max(matched_catalog_stars[:, 2]) + 1
else:
# If there are no matched stars, use the limiting magnitude from config
faintest_mag = config.catalog_mag_limit + 1
# Estimate RA,dec of the centre of the FOV
_, RA_c, dec_c, _ = xyToRaDecPP([jd2Date(max_jd)], [platepar.X_res/2], [platepar.Y_res/2], [1],
platepar)
RA_c = RA_c[0]
dec_c = dec_c[0]
fov_radius = np.hypot(*computeFOVSize(platepar))
# Get stars from the catalog around the defined center in a given radius
_, extracted_catalog = subsetCatalog(catalog_stars, RA_c, dec_c, fov_radius, faintest_mag)
ra_catalog, dec_catalog, mag_catalog = extracted_catalog.T
# Compute image positions of all catalog stars that should be on the image
x_catalog, y_catalog = raDecToXYPP(ra_catalog, dec_catalog, max_jd, platepar)
# Filter all catalog stars outside the image
temp_arr = np.c_[x_catalog, y_catalog, mag_catalog]
temp_arr = temp_arr[temp_arr[:, 0] >= 0]
temp_arr = temp_arr[temp_arr[:, 0] <= ff.avepixel.shape[1]]
temp_arr = temp_arr[temp_arr[:, 1] >= 0]
temp_arr = temp_arr[temp_arr[:, 1] <= ff.avepixel.shape[0]]
x_catalog, y_catalog, mag_catalog = temp_arr.T
# Plot catalog stars on the image
cat_stars_handle = plt.scatter(x_catalog, y_catalog, c='none', marker='D', lw=1.0, alpha=0.4, \
s=((4.0 + (faintest_mag - mag_catalog))/3.0)**(2*2.512), edgecolor='r', label='Catalog stars')
legend_handles.append(cat_stars_handle)
### ###
# Add info text
info_text = ff_dict[max_jd] + '\n' \
+ "Matched stars: {:d}/{:d}\n".format(max_matched_stars, len(star_dict[max_jd])) \
+ "Median distance: {:.2f} px\n".format(np.median(distances)) \
+ "Catalog limiting magnitude: {:.1f}".format(lim_mag)
plt.text(10, 10, info_text, bbox=dict(facecolor='black', alpha=0.5), va='top', ha='left', fontsize=4, \
color='w')
legend = plt.legend(handles=legend_handles, prop={'size': 4}, loc='upper right')
legend.get_frame().set_facecolor('k')
legend.get_frame().set_edgecolor('k')
for txt in legend.get_texts():
txt.set_color('w')
plt.axis('off')
plt.gca().get_xaxis().set_visible(False)
plt.gca().get_yaxis().set_visible(False)
plt.xlim([0, ff.avepixel.shape[1]])
plt.ylim([ff.avepixel.shape[0], 0])
# Remove the margins
plt.subplots_adjust(left=0, bottom=0, right=1, top=1, wspace=0, hspace=0)
plt.savefig(os.path.join(night_dir_path, night_name + '_calib_report_astrometry.jpg'), \
bbox_inches='tight', pad_inches=0, dpi=dpi)
if show_graphs:
plt.show()
else:
plt.clf()
plt.close()
if max_matched_stars > 2:
### Plot the photometry ###
plt.figure(dpi=dpi)
# Take only those stars which are inside the 3/4 of the shorter image axis from the center
photom_selection_radius = np.min([img_h, img_w])/3
filter_indices = ((image_stars[:, 0] - img_h/2)**2 + (image_stars[:, 1] \
- img_w/2)**2) <= photom_selection_radius**2
star_intensities = image_stars[filter_indices, 2]
catalog_mags = matched_catalog_stars[filter_indices, 2]
# Plot intensities of image stars
#star_intensities = image_stars[:, 2]
plt.scatter(-2.5*np.log10(star_intensities), catalog_mags, s=5, c='r')
# Fit the photometry on automated star intensities
photom_offset, fit_stddev, _ = photometryFit(np.log10(star_intensities), catalog_mags)
# Plot photometric offset from the platepar
x_min, x_max = plt.gca().get_xlim()
y_min, y_max = plt.gca().get_ylim()
x_min_w = x_min - 3
x_max_w = x_max + 3
y_min_w = y_min - 3
y_max_w = y_max + 3
photometry_info = 'Platepar: {:+.2f}LSP {:+.2f} +/- {:.2f} \nGamma = {:.2f}'.format(platepar.mag_0, \
platepar.mag_lev, platepar.mag_lev_stddev, platepar.gamma)
# Plot the photometry calibration from the platepar
logsum_arr = np.linspace(x_min_w, x_max_w, 10)
plt.plot(logsum_arr, logsum_arr + platepar.mag_lev, label=photometry_info, linestyle='--', \
color='k', alpha=0.5)
# Plot the fitted photometry calibration
fit_info = "Fit: {:+.2f}LSP {:+.2f} +/- {:.2f}".format(-2.5, photom_offset, fit_stddev)
plt.plot(logsum_arr, logsum_arr + photom_offset, label=fit_info, linestyle='--', color='red',
alpha=0.5)
plt.legend()
plt.ylabel("Catalog magnitude ({:s})".format(mag_band_str))
plt.xlabel("Uncalibrated magnitude")
# Set wider axis limits
plt.xlim(x_min_w, x_max_w)
plt.ylim(y_min_w, y_max_w)
plt.gca().invert_yaxis()
plt.gca().invert_xaxis()
plt.grid()
plt.savefig(os.path.join(night_dir_path, night_name + '_calib_report_photometry.png'), dpi=150)
if show_graphs:
plt.show()
else:
plt.clf()
plt.close()