friendblend.processing package¶

Subpackages¶

Submodules¶

friendblend.processing.alpha_blending module¶

Alpha Blending

friendblend.processing.alpha_blending.alpha_blend(img_l, img_r, bb_l, bb_r)[source]¶: Performs alpha blending on the images

friendblend.processing.color_correction module¶

Contains code for color correction of the image. CLAHE - Contrast Limited Adaptive Histogram Equalization

friendblend.processing.color_correction.apply_clahe(img, clip_limit=3.0, n_bins=256, grid=(7, 7))[source]¶

Parameters:

The input image in BGR format
The clip limit for contrast limiting
Number of bins to divide the image into
Grid size to be used while performing adaptive histogram equalization

friendblend.processing.color_correction.clahe(img, clip_limit=3.0, n_bins=256, grid=(7, 7))[source]¶

Divides the image into frames/cells using the grid values passed to the function
Creates a separate histogram for each cell/frame after binning the values
The histogram equalization mapping for each cell/frame is also generated
The contrast of the image is limited by clipping bin values with ‘clip_limit’
Adaptive histogram equalization

friendblend.processing.color_correction.rgb2lab(img)[source]¶: Converts an RGB image to L*a*b* space Implements formulas described at: https://docs.opencv.org/3.4/de/d25/imgproc_color_conversions.html

friendblend.processing.face_body_detection module¶

Face and body detection

friendblend.processing.face_body_detection.get_body(img, face_bounds)[source]¶: Calculates the body bounds using face bounds

friendblend.processing.face_body_detection.get_bounds(img)[source]¶: Calculates and returns face and body bounds Can die in between

friendblend.processing.face_body_detection.get_face(img)[source]¶: Returns the face bounds Can die in between

friendblend.processing.grab_cut module¶

Grab Cut

friendblend.processing.grab_cut.crop_fg(fg, bg)[source]¶: Superimposes cropped foreground from grabcut onto background image

friendblend.processing.grab_cut.filter_mask(mask: numpy.ndarray, fb: Tuple[int, int, int, int]) → numpy.ndarray[source]¶: Removes components which are not connected to the face bounding box from foreground

friendblend.processing.grab_cut.grab_cut(img_l, img_r, fb_l, boundary=20)[source]¶

Performs grabcut

extracts face from img_l using face bounding boxes
blends the two images

friendblend.processing.helpers module¶

Common functions used multiple times

class friendblend.processing.helpers.DSU(n)[source]¶

Bases: object

The union-find algorithm

find(i)[source]¶: Finds the parent of i’th index

union(i, j)[source]¶: Puts i and j under the same subtree

friendblend.processing.helpers.connected(img: numpy.ndarray) → numpy.ndarray[source]¶: Connected components, implementing 4-connectivity

friendblend.processing.helpers.draw_box(img, bounding_box)[source]¶: Draws a bounding box in image

friendblend.processing.helpers.pt_in_box(pt, box)[source]¶

pt: (x1, y1) box: (x, y, w, h)

returns whether point is within box

friendblend.processing.keypoint module¶

Contains code for keypoint detection and matching

class friendblend.processing.keypoint.ORB(**kwargs)[source]¶