Revision history [back]

Removing image background from image with python

I am using opencv with python for removing background from image. code i have write is working for some image not for all. please help me to find exect solution.

Attaching some sample images : C:\fakepath\ashok.jpg

This is python code :

import cv2 import argparse import numpy as np

parser = argparse.ArgumentParser() parser.add_argument('file_in', help='Input file') parser.add_argument('file_out', help='Output file') args = parser.parse_args()

== Parameters =======================================================================

BLUR = 19 CANNY_THRESH_1 = 25 CANNY_THRESH_2 = 255 MASK_DILATE_ITER = 5 MASK_ERODE_ITER = 13 MASK_COLOR = (0.0,0.0,1.0)

-- Read image -----------------------------------------------------------------------

img = cv2.imread(args.file_in) gray = cv2.GaussianBlur(img, (5, 5), 1) gray = cv2.cvtColor(gray,cv2.COLOR_BGR2RGB) gray = cv2.cvtColor(gray,cv2.COLOR_BGR2GRAY)

gray = 255 - cv2.threshold(gray, 0,255, cv2.THRESH_BINARY+cv2.THRESH_OTSU)[1]

edges = cv2.dilate(gray, kernel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE, (21,21)) , iterations = 1) edges = cv2.morphologyEx(edges, cv2.MORPH_CLOSE, cv2.getStructuringElement(cv2.MORPH_ELLIPSE, (3,3))) edges = cv2.GaussianBlur(edges, (1, 1), 0)

-- Find contours in edges, sort by area ---------------------------------------------

contour_info = [] _, contours, _ = cv2.findContours(edges, cv2.RETR_LIST, cv2.CHAIN_APPROX_NONE)

for c in contours: contour_info.append(( c, cv2.isContourConvex(c), cv2.contourArea(c), )) contour_info = sorted(contour_info, key=lambda c: c[2], reverse=True) max_contour = contour_info[0]

-- Create empty mask, draw filled polygon on it corresponding to largest contour ----

Mask is black, polygon is white

mask = np.zeros(edges.shape) cv2.fillConvexPoly(mask, max_contour[0], (255))

-- Smooth mask, then blur it --------------------------------------------------------

mask = cv2.dilate(mask, None, iterations=MASK_DILATE_ITER)

mask = cv2.GaussianBlur(mask, (BLUR, BLUR), 0)

mask = cv2.erode(mask, None, iterations=MASK_ERODE_ITER)

-- Blend masked img into MASK_COLOR background --------------------------------------

img = img.astype('float32') / 255.0 # for easy blending

cv2.imwrite('mask.png', mask)

c_red, c_green, c_blue = cv2.split(img) img_a = cv2.merge((c_red, c_green, c_blue, mask.astype('float32') / 255.0))

cv2.imwrite(args.file_out, img_a*255)

Thanks Manoj kumar

Removing image background from image with python

I am using opencv with python for removing background from image. code i have write is working for some image not for all. please help me to find exect solution.

Attaching some sample images : C:\fakepath\ashok.jpg

This is python code : : import cv2 import argparse import numpy as np np parser = argparse.ArgumentParser() parser.add_argument('file_in', help='Input file') parser.add_argument('file_out', help='Output file') args = parser.parse_args() == parser.parse_args() #== Parameters ======================================================================= ======================================================================= BLUR = 19 CANNY_THRESH_1 = 25 CANNY_THRESH_2 = 255 MASK_DILATE_ITER = 5 MASK_ERODE_ITER = 13 MASK_COLOR = (0.0,0.0,1.0) -- (0.0,0.0,1.0) #-- Read image ----------------------------------------------------------------------- ----------------------------------------------------------------------- img = cv2.imread(args.file_in) gray = cv2.GaussianBlur(img, (5, 5), 1) gray = cv2.cvtColor(gray,cv2.COLOR_BGR2RGB) gray = cv2.cvtColor(gray,cv2.COLOR_BGR2GRAY) cv2.cvtColor(gray,cv2.COLOR_BGR2GRAY) gray = 255 - cv2.threshold(gray, 0,255, cv2.THRESH_BINARY+cv2.THRESH_OTSU)[1] cv2.THRESH_BINARY+cv2.THRESH_OTSU)[1] edges = cv2.dilate(gray, kernel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE, (21,21)) , iterations = 1) edges = cv2.morphologyEx(edges, cv2.MORPH_CLOSE, cv2.getStructuringElement(cv2.MORPH_ELLIPSE, (3,3))) edges = cv2.GaussianBlur(edges, (1, 1), 0) -- 0) #-- Find contours in edges, sort by area --------------------------------------------- --------------------------------------------- contour_info = [] _, contours, _ = cv2.findContours(edges, cv2.RETR_LIST, cv2.CHAIN_APPROX_NONE) cv2.CHAIN_APPROX_NONE) for c in contours: contour_info.append(( c, cv2.isContourConvex(c), cv2.contourArea(c), )) contour_info = sorted(contour_info, key=lambda c: c[2], reverse=True) max_contour = contour_info[0] -- contour_info[0] #-- Create empty mask, draw filled polygon on it corresponding to largest contour ---- ---- # Mask is black, polygon is white white mask = np.zeros(edges.shape) cv2.fillConvexPoly(mask, max_contour[0], (255)) -- (255)) #-- Smooth mask, then blur it -------------------------------------------------------- mask -------------------------------------------------------- #mask = cv2.dilate(mask, None, iterations=MASK_DILATE_ITER) iterations=MASK_DILATE_ITER) mask = cv2.GaussianBlur(mask, (BLUR, BLUR), 0) mask 0) #mask = cv2.erode(mask, None, iterations=MASK_ERODE_ITER) -- iterations=MASK_ERODE_ITER) #-- Blend masked img into MASK_COLOR background -------------------------------------- -------------------------------------- img = img.astype('float32') / 255.0 # for easy blending cv2.imwrite('mask.png', mask) blending #cv2.imwrite('mask.png', mask) c_red, c_green, c_blue = cv2.split(img) img_a = cv2.merge((c_red, c_green, c_blue, mask.astype('float32') / 255.0)) 255.0)) cv2.imwrite(args.file_out, img_a*255)img_a*255)

Thanks Manoj kumar