Segmentation¶

class segmentation.Segmentation(image, imager, time)[source]¶

Contains all the methods needed for the segmentation a sky/cloud image

image¶: np.array – The image to segment

imager¶: Imager – The imager used to take the image

time¶: datetime.datetime – The time at which the image was captured (UTC+8)

cmask(index, radius, array)[source]¶

Generates the mask for a given input image. The generated mask is needed to remove occlusions during post-processing steps.

Parameters:	index (numpy array) – Array containing the x- and y- co-ordinate of the center of the circular mask. radius (float) – Radius of the circular mask. array (numpy array) – Input sky/cloud image for which the mask is generated.
Returns:	Generated mask image.
Return type:	numpy array

getBRchannel(input_image, mask_image)[source]¶

Extracts the ratio of red and blue blue channel from an input sky/cloud image. It is used in the clustering step to generate the binary sky/cloud image.

Parameters:	input_image (numpy array) – Input sky/cloud image. mask_image (numpy array) – Mask to remove occlusions from the input image. This mask contains boolean values indicating the allowable pixels from an image.
Returns:	Ratio image using red and blue color channels, normalized to [0,255].
Return type:	numpy array

make_cluster_mask(input_matrix, mask_image)[source]¶

Clusters an input sky/cloud image to generate the binary image and the coverage ratio value.

Parameters:	input_matrix (numpy array) – Input matrix that needs to be normalized. mask_image (numpy array) – Mask to remove occlusions from the input image. This mask contains boolean values indicating the allowable pixels from an image.
Returns:	Binary output image, where white denotes cloud pixels and black denotes sky pixels. float: Cloud coverage ratio in the input sky/cloud image. float: The first (out of two) cluster center. float: The second (out of two) cluster center.
Return type:	numpy array

removeSunIfNonCloudy(im_mask)[source]¶

Removes the circumsolar area from the mask if it is detected as non cloudy.

Parameters:	im_mask (numpy array) – Segmentation mask
Returns:	Modified segmentation mask.
Return type:	numpy array

segment()[source]¶: Main function for sky/cloud segmentation. Perform all the necessary steps.

showasImage(input_matrix)[source]¶

Normalizes an input matrix to the range [0,255]. It is useful in displaying the matrix as an image.

Parameters:	input_matrix (numpy array) – Input matrix that needs to be normalized.
Returns:	Returns the normalized matrix.
Return type:	numpy array