Coverage for python / lsst / scarlet / lite / measure.py: 10%

1# This file is part of scarlet_lite. 

2# 

3# Developed for the LSST Data Management System. 

4# This product includes software developed by the LSST Project 

5# (https://www.lsst.org). 

6# See the COPYRIGHT file at the top-level directory of this distribution 

7# for details of code ownership. 

8# 

9# This program is free software: you can redistribute it and/or modify 

10# it under the terms of the GNU General Public License as published by 

11# the Free Software Foundation, either version 3 of the License, or 

12# (at your option) any later version. 

13# 

14# This program is distributed in the hope that it will be useful, 

15# but WITHOUT ANY WARRANTY; without even the implied warranty of 

16# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the 

17# GNU General Public License for more details. 

18# 

19# You should have received a copy of the GNU General Public License 

20# along with this program. If not, see <https://www.gnu.org/licenses/>. 

21 

22from typing import cast 

23 

24import numpy as np 

25 

26from .bbox import Box 

27from .image import Image 

28 

29 

30def calculate_snr( 

31 images: Image, 

32 variance: Image, 

33 psfs: np.ndarray, 

34 center: tuple[int, int], 

35) -> float: 

36 """Calculate the signal to noise for a point source 

37 

38 This is done by weighting the image with the PSF in each band 

39 and dividing by the PSF weighted variance. 

40 

41 Parameters 

42 ---------- 

43 images: 

44 The 3D (bands, y, x) image containing the data. 

45 variance: 

46 The variance of `images`. 

47 psfs: 

48 The PSF in each band. 

49 center: 

50 The center of the signal. 

51 

52 Returns 

53 ------- 

54 snr: 

55 The signal to noise of the source. 

56 """ 

57 py = psfs.shape[1] // 2 

58 px = psfs.shape[2] // 2 

59 bbox = Box(psfs[0].shape, origin=(-py + center[0], -px + center[1])) 

60 overlap = images.bbox & bbox 

61 noise = variance[overlap].data 

62 img = images[overlap].data 

63 _psfs = Image(psfs, bands=images.bands, yx0=cast(tuple[int, int], bbox.origin))[overlap].data 

64 numerator = img * _psfs 

65 denominator = (_psfs * noise) * _psfs 

66 return np.sum(numerator) / np.sqrt(np.sum(denominator)) 

67 

68 

69def conserve_flux(blend, mask_footprint: bool = True, images: Image | None = None) -> None: 

70 """Use the source models as templates to re-distribute flux 

71 from the data 

72 

73 The source models are used as approximations to the data, 

74 which redistribute the flux in the data according to the 

75 ratio of the models for each source. 

76 There is no return value for this function, 

77 instead it adds (or modifies) a ``flux_weighted_image`` 

78 attribute to each the sources with the flux attributed to 

79 that source. 

80 

81 Parameters 

82 ---------- 

83 blend: 

84 The blend that is being fit 

85 mask_footprint: 

86 Whether or not to apply a mask for pixels with zero weight. 

87 """ 

88 observation = blend.observation 

89 py = observation.psfs.shape[-2] // 2 

90 px = observation.psfs.shape[-1] // 2 

91 

92 if images is None: 

93 images = observation.images.copy() 

94 if mask_footprint: 

95 images.data[observation.weights.data == 0] = 0 

96 model = blend.get_model() 

97 bands = None 

98 else: 

99 bands = images.bands 

100 model = blend.get_model()[bands,] 

101 # Always convolve in real space to avoid FFT artifacts 

102 model = observation.convolve(model, mode="real") 

103 model.data[model.data < 0] = 0 

104 

105 for src in blend.sources: 

106 if src.is_null: 

107 src.flux_weighted_image = Image.from_box(Box((0, 0)), bands=observation.bands) # type: ignore 

108 continue 

109 src_model = src.get_model() 

110 

111 # Grow the model to include the wings of the PSF 

112 src_box = src.bbox.grow((py, px)) 

113 overlap = observation.bbox & src_box 

114 src_model = src_model.project(bbox=overlap) 

115 src_model = observation.convolve(src_model, mode="real") 

116 if bands is not None: 

117 src_model = src_model[bands,] 

118 src_model.data[src_model.data < 0] = 0 

119 numerator = src_model.data 

120 denominator = model[overlap].data 

121 cuts = denominator != 0 

122 ratio = np.zeros(numerator.shape, dtype=numerator.dtype) 

123 ratio[cuts] = numerator[cuts] / denominator[cuts] 

124 ratio[denominator == 0] = 0 

125 # sometimes numerical errors can cause a hot pixel to have a 

126 # slightly higher ratio than 1 

127 ratio[ratio > 1] = 1 

128 src.flux_weighted_image = src_model.copy_with(data=ratio) * images[overlap]