pairwisehelper

Helper functions for the class pairwise alignment Creates its own logger object when imported.

Functions:

pysmFISH.stitching_package.pairwisehelper.calc_translated_pics(trans, overlap1, overlap2, round_size=False)

Calculate wich part of 2 pictures overlap after translation.

Translates overlap2 and returns the parts of overlap1 and overlap2 that overlap after translation. Parts that do not overlap are cropped. The alternative would be to fill the parts that do not overlap with zeros, but this seems to mess up the covariance calculation more than cropping the pictures. Here, one problem is that we may crop unnecessarily much because the cropping is done on the overlap and not on the whole tile.

trans: np.array

1 by 2 np-array with y and x translation.

overlap1: np.array

Image

overlap2: np.array

Image that is supposed to overlap with overlap1

round_size: bool

If True the final image sizes will be rounded to the nearest integer before warping the images. (Default: False)

shifted_a: np.array

Overlap a shifted and cropped to the same size as shifted_b

shifted_b: np.array

Overlap b shifted and cropped to the same size as shifted_a

pysmFISH.stitching_package.pairwisehelper.calc_translated_pics_3D(trans, overlap1, overlap2, round_size=False)

Calculate wich part of two 3D pictures overlap after translation.

Translates overlap2 and returns the parts of overlap1 and overlap2 that overlap after translation. This function works in 3D but does not translate in z. Parts that do not overlap are cropped. The alternative would be to fill the parts that do not overlap with zeros, but this seems to mess up the covariance calculation more than cropping the pictures. Here, one problem is that we may crop unnecessarily much because the cropping is done on the overlap and not on the whole tile.

trans: np.array

1 by 2 np-array with y and x translation.

overlap1: np.array

Image

overlap2: np.array

Image that is supposed to overlap with overlap1

round_size: bool

If True the final image sizes will be rounded to the nearest integer before warping the images. (Default: False)

shifted_a: np.array

Overlap a shifted and cropped to the same size as shifted_b

shifted_b: np.array

Overlap b shifted and cropped to the same size as shifted_a

pysmFISH.stitching_package.pairwisehelper.calculate_PCM(pic_a, pic_b)

Calculate the phase correlation matrix of pic_a and pic_b.

pic_a: np.array

First picture

pic_b: np.array

Second picture

r: np.array

Normalized PCM matrix

pic_a and pic_b should have the same size

pysmFISH.stitching_package.pairwisehelper.calculate_PCM_method2(pic_a, pic_b)

Calculate the phase correlation matrix of pic_a and pic_b.

Method 2 differs from calculate_PCM in that it calculates the PCM per, layer in the image therefore this function is only applicable to 3D images/complete z-stacks. In fact it is only used when in pairwisesingle.py if method == ‘calculate per layer’.

pic_a: np.array

First picture

pic_b: np.array

Second picture

r: np.array

Normalized PCM matrix

pic_a and pic_b should have the same size

pysmFISH.stitching_package.pairwisehelper.display_overlap(overlap1, overlap2, best_trans, best_cov, plot_order)

Plot two pictures and show how they overlap.

This function will only plot when allow_plot is True and in inout matplotlib is imported and plot_available is True.

overlap1: np.array

2d np-array representing an image

overlap2: np.array

2d np-array representing an image

best_trans: np.array

1 by 2 np array with an y and x value for the best transition

best_cov: float

Number indicating the goodness of the overlap between the two pictures

plot_order: list

The order in wich subplots should be placed if when plotting overlap1 and 2 and overlap rgb

pysmFISH.stitching_package.pairwisehelper.get_overlapping_region(tile_1, tile_2, overlap_ind_x, overlap_ind_y, direction)

Given the overlap indexes calculate the overlap in direction.

tile_1: np.array

2D or 3D numpy array representing an image.

tile_2: np.array

2D or 3D numpy array representing an image.

overlap_ind_x: int

The difference in pixels between the tile corners in the x/column direction.

overlap_ind_y: int

The difference in pixels between the tile corners in the y/row direction.

direction: str

Valid values: ‘left’ or ‘top’. Denotes if the two tiles neighbour on the right and left or at the bottom and top.

overlap_1: np.array

2D or 3D numpy arrays. The parts of each image that overlap.

overlap_2: np.array

2D or 3D numpy arrays. The parts of each image that overlap.

pysmFISH.stitching_package.pairwisehelper.xcov_nd(pic_1, pic_2)

Calculate the normalized cross covariance of pic_1 and pic_2.

pic_1: np.array

Numpy array representing a gray value picture, same size as pic_2

pic_2: np.array

Numpy array representing a gray value picture, same size as pic_1

: float

The normalized crosscovariance of pic_1 and pic_2 (coVar / (stDev1 * stDev2))

monocolor: bool

True if the one or both of the pictures contained only one color.