[ { "path": "Centroid/Centroid.py", "content": "import cv2\n\nimg = cv2.imread(\"signature.png\", 0)\n\nret,binary = cv2.threshold(img,10,255,cv2.THRESH_BINARY)\n\ndef findBB(im):\n h, w = im.shape[0], im.shape[1] \n left, top = w, h\n right, bottom = 0, 0\n \n for x in xrange(h):\n for y in xrange(w):\n if (im[x,y] == 0):\n right = x if x > right else right\n left = x if x < left else left\n bottom = y if y > bottom else bottom\n top = y if y < top else top\n \n return (left, right, top, bottom)\n\ndef findCentroid(im):\n h, w = im.shape[0], im.shape[1]\n cx, cy, n = 0, 0, 0\n for x in xrange(h):\n for y in xrange(w):\n if (im[x,y] == 0):\n cx += x\n cy += y\n n += 1\n cx /= n\n cy /= n\n return (cx, cy)\n\ndef divideImgIntoFour(im, cent):\n h, w = im.shape[0], im.shape[1]\n cx, cy = cent\n img1 = im[0:cx, 0:cy]\n img2 = im[0:cx, cy:w]\n img3 = im[cx:h, 0:cy]\n img4 = im[cx:h, cy:w]\n return [img1, img2, img3, img4]\n\ndef calculateTransitions(im):\n h, w = im.shape[0], im.shape[1]\n prev = im[0,0]\n n = 0\n for x in range(1, h):\n for y in range(1, w):\n curr = im[x,y]\n # check if the is black to white transition\n n = n+1 if curr == 255 and prev == 0 else n\n prev = curr\n return n\n\nboundingBox = findBB(binary)\ncropImg = binary[boundingBox[0]:boundingBox[1], boundingBox[2]:boundingBox[3]]\ncentroid = findCentroid(cropImg)\nsegments = divideImgIntoFour(cropImg, centroid)\ntransitions = [calculateTransitions(seg) for seg in segments]\n\nprint \"Bounding Box:\", boundingBox\nprint \"Coordinates of centroid:\", centroid\nprint \"Black to white transitions (4 segments):\", transitions\n\ncv2.imshow(\"TopLeft\", segments[0])\ncv2.imwrite(\"TopLeft.png\", segments[0])\ncv2.imshow(\"TopRight\", segments[1])\ncv2.imwrite(\"TopRight.png\", segments[1])\ncv2.imshow(\"BottomLeft\", segments[2])\ncv2.imwrite(\"BottomLeft.png\", segments[2])\ncv2.imshow(\"BottomRight\", segments[3])\ncv2.imwrite(\"BottomRight.png\", segments[3])\ncv2.waitKey(0)\n\n" }, { "path": "Connected Component Labelling/ccl4.py", "content": "import Image, ImageOps\nimport sys\nimport random\nimport numpy\n\n\ndef colourize(img):\n height, width = img.shape\n\n colors = []\n colors.append([])\n colors.append([])\n color = 1\n # Displaying distinct components with distinct colors\n coloured_img = Image.new(\"RGB\", (width, height))\n coloured_data = coloured_img.load()\n\n for i in range(len(img)):\n for j in range(len(img[0])):\n if img[i][j] > 0:\n if img[i][j] not in colors[0]:\n colors[0].append(img[i][j])\n colors[1].append((random.randint(0, 255), random.randint(0, 255), random.randint(0, 255)))\n\n ind = colors[0].index(img[i][j])\n coloured_data[j, i] = colors[1][ind]\n\n return coloured_img\n\n\ndef binarize(img_array, threshold=130):\n for i in range(len(img_array)):\n for j in range(len(img_array[0])):\n if img_array[i][j] > threshold:\n img_array[i][j] = 0\n else:\n img_array[i][j] = 1\n return img_array\n\n\ndef ccl4(img_array):\n ##### first pass #####\n print \"starting first pass\"\n curr_label = 1;\n img_array = numpy.array(img_array)\n labels = numpy.array(img_array)\n\n # storing label conversions\n label_conv = []\n label_conv.append([])\n label_conv.append([])\n\n count = 0\n for i in range(1, len(img_array)):\n for j in range(1, len(img_array[0])):\n\n if img_array[i][j] > 0:\n label_x = labels[i][j - 1]\n label_y = labels[i - 1][j]\n\n if label_x > 0:\n # both x and y have a label\n if label_y > 0:\n\n if not label_x == label_y:\n labels[i][j] = min(label_x, label_y)\n if max(label_x, label_y) not in label_conv[0]:\n label_conv[0].append(max(label_x, label_y))\n label_conv[1].append(min(label_x, label_y))\n elif max(label_x, label_y) in label_conv[0]:\n ind = label_conv[0].index(max(label_x, label_y))\n if label_conv[1][ind] > min(label_x, label_y):\n l = label_conv[1][ind]\n label_conv[1][ind] = min(label_x, label_y)\n while l in label_conv[0] and count < 100:\n count += 1\n ind = label_conv[0].index(l)\n l = label_conv[1][ind]\n label_conv[1][ind] = min(label_x, label_y)\n\n label_conv[0].append(l)\n label_conv[1].append(min(label_x, label_y))\n\n else:\n labels[i][j] = label_y\n # only x has a label\n else:\n labels[i][j] = label_x\n\n # only y has a label\n elif label_y > 0:\n labels[i][j] = label_y\n\n # neither x nor y has a label\n else:\n labels[i][j] = curr_label\n curr_label += 1\n\n ##### second pass #####\n print \"starting second pass\"\n count = 1\n for idx, val in enumerate(label_conv[0]):\n\n if label_conv[1][idx] in label_conv[0] and count < 100:\n count += 1\n ind = label_conv[0].index(label_conv[1][idx])\n label_conv[1][idx] = label_conv[1][ind]\n\n for i in range(1, len(labels)):\n for j in range(1, len(labels[0])):\n\n if labels[i][j] in label_conv[0]:\n ind = label_conv[0].index(labels[i][j])\n labels[i][j] = label_conv[1][ind]\n\n return labels\n\n\ndef main():\n numpy.set_printoptions(threshold=numpy.nan)\n # Open the image\n img = Image.open(sys.argv[1])\n\n # Threshold the image\n img = img.convert('L')\n img = ImageOps.expand(img, border=1, fill='white')\n img = numpy.array(img)\n img = binarize(img)\n\n \"\"\"\n img = [ [0,0,0,0,0,0,0,0,0,0],\n [0,1,1,0,1,1,1,0,1,0],\n [0,1,1,0,1,0,1,0,1,0],\n [0,1,1,1,1,0,0,0,1,0],\n [0,0,0,0,0,0,0,0,1,0],\n [0,1,1,1,1,0,1,0,1,0],\n [0,0,0,0,1,0,1,0,1,0],\n [0,1,1,1,1,0,0,0,1,0],\n [0,1,1,1,1,0,1,1,1,0],\n [0,0,0,0,0,0,0,0,0,0]]\n \"\"\"\n\n img = ccl4(img)\n\n # Colour the image using labels\n coloured_img = colourize(img)\n\n # Show the coloured image\n coloured_img.show()\n\n\nif __name__ == \"__main__\": main()\n" }, { "path": "Gradient/gradient.py", "content": "from PIL import Image\r\nimport cv2\r\nimport numpy as np\r\n\r\n#Read Image in GrayScale\r\nimg_gray = cv2.imread('lena.jpg',0) \r\nh,w = img_gray.shape[:2]\r\n\r\ngrad_img = np.asarray(img_gray)\r\n\r\nfor i in range(0,h):\r\n for j in range(0,w-1):\r\n\r\n #applying gradient\r\n\ta = min(img_gray[i][j+1],img_gray[i][j])\r\n\tif a == img_gray[i][j+1] :\r\n \t\ttemp_arr = img_gray[i][j] - img_gray[i][j+1]\r\n\telse :\r\n\t\ttemp_arr = img_gray[i][j+1] - img_gray[i][j]\r\n \r\n \r\n grad_img[i,j] = temp_arr\r\n\r\nimg = Image.fromarray(grad_img) \r\nimg.save(\"gradient.jpg\")\r\n\r\n" }, { "path": "Histogram Equalization/hist_eq.py", "content": "from PIL import Image\nimport matplotlib.pyplot as plt\nimport cv2\nimport numpy as np\n\nimg = cv2.imread('hist2.tif',0)\n\n#Initialize intensity values with 256 zeroes\nintensity_count = [0] * 256 \n\nheight,width = img.shape[:2] \nN = height * width \n\n\t#Array for new_image\nhigh_contrast = np.zeros(img.shape) \n\nfor i in range(0,height):\n \t\tfor j in range(0,width):\n \t\t\tintensity_count[img[i][j]] += 1 #Find pixels count for each intensity\n\nL = 256\n\nintensity_count,total_values_used = np.histogram(img.flatten(),L,[0,L]) \npdf_list = np.ceil(intensity_count*(L-1)/img.size) #Calculate PDF\ncdf_list = pdf_list.cumsum() \t#Calculate CDF\n\n\nfor y in range(0, height):\n \t\tfor x in range(0, width): \n\t\t\t#Apply the new intensities in our new image\n \t\t\thigh_contrast[y,x] = cdf_list[img[y,x]] \n\n\t\n#PLOT THE HISTOGRAMS\ncv2.imwrite('high_contrast.png', high_contrast) \n\nplt.hist(img.ravel(),256,[0,256])\nplt.xlabel('Intensity Values')\nplt.ylabel('Pixel Count')\nplt.show()\n\nplt.hist(high_contrast.ravel(),256,[0,256])\t\nplt.xlabel('Intensity Values')\nplt.ylabel('Pixel Count')\nplt.show()\n" }, { "path": "Image Negative/negative.py", "content": "from PIL import Image\r\nimport cv2\r\nimport sys\r\nimport numpy as np\r\n\r\nS = 255\r\n\r\n# if in rgb\r\nif(sys.argv[2]==\"rgb\"):\r\n # open in rgb\r\n img = cv2.imread(sys.argv[1],cv2.IMREAD_COLOR) \r\n B,G,R = cv2.split(img)\r\n B[:] = [S-x for x in B] #inverting blue\r\n G[:] = [S-x for x in G] #inverting green \r\n R[:] = [S-x for x in R] #inverting red\r\n\r\n #saving image\r\n my_img = cv2.merge((B, G, R)) \r\n cv2.imwrite(sys.argv[1]+'_inverted.png', my_img)\r\n cv2.imshow(sys.argv[1]+'_inverted.png', my_img) \r\n\r\n#if in grayscale or binary\r\nelse: \r\n # open in grayscale\r\n img = cv2.imread(sys.argv[1],cv2.IMREAD_GRAYSCALE) \r\n my_img = np.array([S-x for x in img])\r\n cv2.imwrite(sys.argv[1]+'_inverted.png', my_img)\r\n cv2.imshow(sys.argv[1]+'_inverted.png', my_img) \r\n \r\n\r\n\r\n\r\n\r\n\r\n" }, { "path": "Image Segmentation/Segmentation.py", "content": "import cv2\nimport numpy as np\n\n# Loading the image in RGB\nimg = cv2.imread(\"image.png\", 0)\n\n# Applying Gaussian blur with kernel size 7 to remove unwanted noise\nblurred_image = cv2.GaussianBlur(img,(7,7),0)\n\n# Applying Otsu's thresholding to binarize the image\nretval ,binarized_image = cv2.threshold(blurred_image,40,255,cv2.THRESH_BINARY)\n\n# Applying Closing to fill in the holes\nfilter = np.ones((3,3),np.uint8)\nclosed_image = cv2.morphologyEx(binarized_image, cv2.MORPH_CLOSE, filter)\n\n# Using connected components to label the image\nretval, markers = cv2.connectedComponents(closed_image)\n\n# Mapping the component labels to hue val\nlabel_hue = np.uint8(120*markers/np.max(markers))\nblank_ch = 255*np.ones_like(label_hue)\nlabeled_image = cv2.merge([label_hue, blank_ch, blank_ch])\n\n# changing from HSV to RGB again to show\nlabeled_image = cv2.cvtColor(labeled_image, cv2.COLOR_HSV2BGR)\n\n# background label set to black\nlabeled_image[label_hue==0] = 0\n\n# getting the unique colors in the image\nunique_colors = np.unique(labeled_image.reshape(-1, labeled_image.shape[2]), axis=0)\n\nprint \"Colors available in labeled image:\"\nfor x in xrange(unique_colors.shape[0]):\n print str(x+1)+\"=> B:\"+str(unique_colors[x,0])+\" G:\"+str(unique_colors[x,1])+\" R:\"+str(unique_colors[x,2])+\" \"\n\nprint \"\\nSelect one of the colors and give its RGB values \"\n\nr = raw_input(\"B : \")\ng = raw_input(\"G : \")\nb = raw_input(\"R : \")\n\n# making an output image\noutput_image = np.zeros_like(labeled_image)\n\n# getting the object of user input color\nfor x in xrange(labeled_image.shape[0]):\n for y in xrange(labeled_image.shape[1]):\n if (labeled_image[x,y,0] == int(r) and labeled_image[x,y,1] == int(g) and labeled_image[x,y,2] == int(b)):\n output_image[x,y,0:3] = labeled_image[x,y,0:3]\n\n# show the output image\ncv2.imshow(\"Selected\", labeled_image)\ncv2.waitKey(0)\n" }, { "path": "Local Histogram Analysis/Slidingwindow.py", "content": "import cv2\nimport numpy as np\nimport matplotlib.pyplot as plt\nplt.figure(figsize=(12,12))\n\n#reading image as grayscale\nimg = cv2.imread(\"mountains.jpg\",0)\n\ndef slidingWindowEqualization(im, winSize):\n newImg = np.zeros((im.shape[0], im.shape[1]))\n for row in xrange(im.shape[0]-winSize+1):\n for col in xrange(im.shape[1]-winSize+1):\n newImg[row:row+winSize,col:col+winSize] = cv2.equalizeHist(im[row:row+winSize,col:col+winSize])\n return newImg\n\nwindowSize = 256\noutput_img = slidingWindowEqualization(img, windowSize)\nplt.subplot(211)\nplt.axis('off')\nplt.title(\"Image after transformation\")\nplt.imshow(output_img, cmap=\"gray\")\n\n#writing output image\ncv2.imwrite(\"SlidingWindow.jpg\", output_img)\n\nplt.subplot(212)\nplt.hist(output_img.ravel(),256,[0,256])\nplt.title(\"Histogram\")\n\nplt.show()\n\n" }, { "path": "Local Histogram Analysis/tilingGlobal.py", "content": "from PIL import Image\nimport matplotlib.pyplot as plt\nimport cv2\nimport numpy as np\n\nimg = cv2.imread('mountains.jpg',0)\nintensity_count = [0] * 256 #Initialize intensity values with 256 zeroes\nL = 255.0\n\nin1 = [0] * 256\nin2 = [0] * 256\nin3 = [0] * 256\nin4 = [0] * 256\n\nheight, width = img.shape[:2] #Get width and height\nN = height * width * 1.0 #Get total Pixels\nhalf_height, half_width = height/2, width/2\nhalf_N = (N / 2) * 1.0\n\nim1 = np.zeros((256,256,0))\nim2 = np.zeros((256,256,0))\nim3 = np.zeros((256,256,0))\nim4 = np.zeros((256,256,0))\nhigh_contrast = np.zeros(img.shape) #Array for global_equalized_new_image\nhigh_contrast_local = np.zeros(img.shape) #Array for local_equalized_new_image\n\nfor y in range(0, height):\n for x in range(0, width): \n intensity_count[img[y,x]] += 1 #Increment each gray_level pixel value according to tile (tile 1, 2, 3 or 4)\n if y <= height/2 and x < width/2:\n in1[img[y,x]] += 1\n elif y <= height/2 and x >= width/2:\n in2[img[y,x]] += 1\n elif y > height/2 and x < width/2:\n in3[img[y,x]] += 1\n elif y > height/2 and x >= width/2:\n in4[img[y,x]] += 1\n\nnewList1 = [x / half_N for x in in1]\nnewList2 = [x / half_N for x in in2]\nnewList3 = [x / half_N for x in in3]\nnewList4 = [x / half_N for x in in4]\n\npdf1_array = np.asarray(newList1); cdf1_array = np.cumsum(pdf1_array)\npdf2_array = np.asarray(newList1); cdf2_array = np.cumsum(pdf2_array)\npdf3_array = np.asarray(newList1); cdf3_array = np.cumsum(pdf3_array)\npdf4_array = np.asarray(newList1); cdf4_array = np.cumsum(pdf4_array)\n\napprox1_List = [round(x * L) for x in cdf1_array]\napprox2_List = [round(x * L) for x in cdf2_array]\napprox3_List = [round(x * L) for x in cdf3_array]\napprox4_List = [round(x * L) for x in cdf4_array]\n\nfor y in range(0, height):\n for x in range(0, width): \n if y <= height/2 and x < width/2:\n high_contrast_local[y,x] = approx1_List[img[y,x]]\n elif y <= height/2 and x >= width/2:\n high_contrast_local[y,x] = approx2_List[img[y,x]]\n elif y > height/2 and x < width/2:\n high_contrast_local[y,x] = approx3_List[img[y,x]]\n elif y > height/2 and x >= width/2:\n high_contrast_local[y,x] = approx4_List[img[y,x]]\n\ncv2.imwrite('high_contrast_local.png', high_contrast_local)\n\nnewList = [x / N for x in intensity_count]\n\npdf_array = np.asarray(newList)\ncdf_array = np.cumsum(pdf_array)\n\napprox_List = [round(x * L) for x in cdf_array]\n\nfor y in range(0,height):\n for x in range(0,width):\n high_contrast[y,x] = approx_List[img[y,x]] \n\nplt.hist(high_contrast_local.ravel(),256,[0,256])\nplt.xlabel('Intensity Values')\nplt.ylabel('Pixel Count')\nplt.savefig('high_contrast_local.png')\nplt.clf()\n#plt.show()\n\n\ncv2.imwrite('high_contrast_global.png', high_contrast) #PLOT THE HISTOGRAMS\nplt.hist(img.ravel(),256,[0,256])\nplt.xlabel('Intensity Values')\nplt.ylabel('Pixel Count')\nplt.savefig('Original.png')\nplt.clf()\n#plt.show()\nplt.hist(high_contrast.ravel(),256,[0,256])\nplt.xlabel('Intensity Values')\nplt.ylabel('Pixel Count')\nplt.savefig('high_contrast_global.png')\nplt.clf()\n#plt.show()\n\n\n" }, { "path": "Morphology/segment.py", "content": "import cv2\nimport numpy as np\n\nimg = cv2.imread('inp.jpg',0)\n\nret,bin = cv2.threshold(img,127,255,cv2.THRESH_BINARY_INV)\n\nkernel = np.ones((3,3),np.uint8)\nopened = cv2.morphologyEx(bin, cv2.MORPH_OPEN, kernel)\ncv2.imshow(\"opened\", opened)\n\nkernel = np.ones((5,5),np.uint8)\nclosed = cv2.morphologyEx(opened, cv2.MORPH_CLOSE, kernel)\nret, output = cv2.threshold(closed,127,255,cv2.THRESH_BINARY_INV)\ncv2.imshow(\"segment\", output)\ncv2.imwrite(\"segment.png\", output)\ncv2.waitKey(0)\n" }, { "path": "Morphology/simple.py", "content": "import cv2\nimport numpy as np\n\nimg = cv2.imread('signature.png', 0)\nr, img = cv2.threshold(img, 130, 255, cv2.THRESH_BINARY_INV)\ncv2.imshow(\"Original\", img)\ncv2.waitKey(0)\n\nkernel = np.ones((5,5),np.uint8)\n\ndilated = cv2.dilate(img,kernel,iterations = 1)\ncv2.imshow(\"Dilation\", dilated)\ncv2.imwrite(\"dilation.png\", dilated)\ncv2.waitKey(0)\n\neroded = cv2.erode(img,kernel,iterations = 1)\ncv2.imshow(\"Erosion\", eroded)\ncv2.imwrite(\"erosion.png\", eroded)\ncv2.waitKey(0)\n\nopened = cv2.morphologyEx(img, cv2.MORPH_OPEN, kernel)\ncv2.imshow(\"Opening\", opened)\ncv2.imwrite(\"opening.png\", opened)\ncv2.waitKey(0)\n\nclosed = cv2.morphologyEx(img, cv2.MORPH_CLOSE, kernel)\ncv2.imshow(\"Closing\", closed)\ncv2.imwrite(\"closing.png\", closed)\ncv2.waitKey(0)\n" }, { "path": "README.md", "content": "# Basic Digital Image Processing Tasks\n> This repository contains basic implementations of image processing algorithms in python.\n\n## Required Libraries\n*\tPIL\n```shell\n$ pip install pillow\n```\n*\topencv-python\n```shell\n$ pip install opencv-python\n```\n\n## Algorithms\n\n### Gradient\n\n```shell\n$ python gradient.py\n```\n|Original|Gradient|\n|---|---|\n|![Gradient-Original](https://github.com/mohammaduzair9/Basic-Digital-Image-Processing/blob/master/Gradient/lena.jpg)|![Gradient-Result](https://github.com/mohammaduzair9/Basic-Digital-Image-Processing/blob/master/Gradient/gradient.jpg)|\n\n### Image Negative\n\n```shell\n$ python negative.py binary.jpeg binary\n```\n|Original|Binary Negative|\n|---|---|\n|![Binary-Negative-Original](https://github.com/mohammaduzair9/Basic-Digital-Image-Processing/blob/master/Image%20Negative/binary.jpg)|![Binary-Negative-Result](https://github.com/mohammaduzair9/Basic-Digital-Image-Processing/blob/master/Image%20Negative/binary_inverted.png)|\n\n```shell\n$ python negative.py lena.jpg gray\n```\n|Original|Grayscale Negative|\n|---|---|\n|![Gray-Negative-Original](https://github.com/mohammaduzair9/Basic-Digital-Image-Processing/blob/master/Image%20Negative/grayscale.png)|![Gray-Negative-Result](https://github.com/mohammaduzair9/Basic-Digital-Image-Processing/blob/master/Image%20Negative/grayscale_inverted.png)|\n\n\n```shell\n$ python negative.py lena.jpg rgb\n```\n|Original|RGB Negative|\n|---|---|\n|![Rgb-Negative-Original](https://github.com/mohammaduzair9/Basic-Digital-Image-Processing/blob/master/Image%20Negative/rgb.jpg)|![Rgb-Negative-Result](https://github.com/mohammaduzair9/Basic-Digital-Image-Processing/blob/master/Image%20Negative/rgb_inverted.png)|\n\n### Image Segmentation\n\n```shell\n$ python Segmentation.py\n```\n|Original|Segmented|\n|---|---|\n|![Segmented-Original](https://github.com/mohammaduzair9/Basic-Digital-Image-Processing/blob/master/Image%20Segmentation/image.png)|![Segmented-Result](https://github.com/mohammaduzair9/Basic-Digital-Image-Processing/blob/master/Image%20Segmentation/Capture3.PNG)|\n\n### Centroid\n\n```shell\n$ python Centroid.py\n```\n|Original|Centroid|\n|---|---|\n|![Centroid-Original](https://github.com/mohammaduzair9/Basic-Digital-Image-Processing/blob/master/Centroid/Signature.png)|
Top LeftTop Right
![Centroid-TopLeft](https://github.com/mohammaduzair9/Basic-Digital-Image-Processing/blob/master/Centroid/TopLeft.png)![Centroid-TopRight](https://github.com/mohammaduzair9/Basic-Digital-Image-Processing/blob/master/Centroid/TopRight.png)
Bottom LeftBottom Right
![Centroid-BottomLeft](https://github.com/mohammaduzair9/Basic-Digital-Image-Processing/blob/master/Centroid/BottomLeft.png)![Centroid-BottomRight](https://github.com/mohammaduzair9/Basic-Digital-Image-Processing/blob/master/Centroid/BottomRight.png)
|\n\n### Connected Component Labelling\n\n```shell\n$ python ccl4.py\n```\n|Original|CCL4 Labelled|\n|---|---|\n|![CCL4-Original](https://github.com/mohammaduzair9/Basic-Digital-Image-Processing/blob/master/Connected%20Component%20Labelling/input.png)|![CCL4-Result](https://github.com/mohammaduzair9/Basic-Digital-Image-Processing/blob/master/Connected%20Component%20Labelling/ccl.png)|\n\n### Histogram Equalization\n\n```shell\n$ python hist_eq.py\n```\n|Original|Histogram Equalized|\n|---|---|\n|![Hist-eq-Original](https://github.com/mohammaduzair9/Basic-Digital-Image-Processing/blob/master/Histogram%20Equalization/hist2.jpg)|![Hist-eq-Result](https://github.com/mohammaduzair9/Basic-Digital-Image-Processing/blob/master/Histogram%20Equalization/high_contrast.png)|\n\n### Local Histogram Analysis\n\n|Original|Local Histogram|\n|---|---|\n|![Local-Hist-Original](https://github.com/mohammaduzair9/Basic-Digital-Image-Processing/blob/master/Local%20Histogram%20Analysis/mountains.jpg)|![Local-Hist-Result](https://github.com/mohammaduzair9/Basic-Digital-Image-Processing/blob/master/Local%20Histogram%20Analysis/high_contrast_local_img.png)|\n\n### Morphology\n\n```shell\n$ python Simple.py\n```\n|Original|Morphology|\n|---|---|\n|![Morphology-Original](https://github.com/mohammaduzair9/Basic-Digital-Image-Processing/blob/master/Morphology/signature.png)|
ErosionDilation
![Erosion](https://github.com/mohammaduzair9/Basic-Digital-Image-Processing/blob/master/Morphology/erosion.png)![Dilation](https://github.com/mohammaduzair9/Basic-Digital-Image-Processing/blob/master/Morphology/dilation.png)
OpeningClosing
![Opening](https://github.com/mohammaduzair9/Basic-Digital-Image-Processing/blob/master/Morphology/opening.png)![Closing](https://github.com/mohammaduzair9/Basic-Digital-Image-Processing/blob/master/Morphology/closing.png)
|\n\n### Sharpening\n\n```shell\n$ python sharpen.py\n```\n|Original|Sharpened|\n|---|---|\n|![Sharpened-Original](https://github.com/mohammaduzair9/Basic-Digital-Image-Processing/blob/master/Sharpening/inp1.jpg)|![Sharpened-Result](https://github.com/mohammaduzair9/Basic-Digital-Image-Processing/blob/master/Sharpening/sharpen.jpg)|\n\n### Skeletonization\n\n```shell\n$ python Skeletonization.py\n```\n![Skeletionization](https://github.com/mohammaduzair9/Basic-Digital-Image-Processing/blob/master/Skeletonization/output.png)\n\n### Smoothing\n\n```shell\n$ python AvergingFilter.py\n```\n|Original|Averaging Filter|\n|---|---|\n|![Averaging-Original](https://github.com/mohammaduzair9/Basic-Digital-Image-Processing/blob/master/Smoothing/inp1.jpeg)|![Averaging-Result](https://github.com/mohammaduzair9/Basic-Digital-Image-Processing/blob/master/Smoothing/averaging.jpg)|\n\n```shell\n$ python gaussian.py\n```\n|Original|Gaussian|\n|---|---|\n|![gaussian-Original](https://github.com/mohammaduzair9/Basic-Digital-Image-Processing/blob/master/Smoothing/inp1.jpeg)|![gaussian-Result](https://github.com/mohammaduzair9/Basic-Digital-Image-Processing/blob/master/Smoothing/gaussian.jpg)|\n\n```shell\n$ python unsharp_masking.py\n```\n|Original|Unsharp Masking|\n|---|---|\n|![Unsharp-Original](https://github.com/mohammaduzair9/Basic-Digital-Image-Processing/blob/master/Smoothing/inp2.jpeg)|![Unsharp-Result](https://github.com/mohammaduzair9/Basic-Digital-Image-Processing/blob/master/Smoothing/unsharp_masking.jpg)|\n\n```shell\n$ python median.py\n```\n|Original|Median|\n|---|---|\n|![Unsharp-Original](https://github.com/mohammaduzair9/Basic-Digital-Image-Processing/blob/master/Smoothing/inp3.jpeg)|![Unsharp-Result](https://github.com/mohammaduzair9/Basic-Digital-Image-Processing/blob/master/Smoothing/median.jpg)|\n\n\n### XY Cuts\n\n```shell\n$ python XY_Cuts.py\n```\n|Original|XY Cuts|\n|---|---|\n|![XY-Original](https://github.com/mohammaduzair9/Basic-Digital-Image-Processing/blob/master/XY_Cuts/XY-cuts.png)|![XY-Result](https://github.com/mohammaduzair9/Basic-Digital-Image-Processing/blob/master/XY_Cuts/xycut.png)|\n\n### Template Matching\n\n```shell\n$ python TemplateMatching.py\n```\n|Template|Matched in Image|\n|---|---|\n|![Template](https://github.com/mohammaduzair9/Basic-Digital-Image-Processing/blob/master/Template%20Matching/template.png)|![MatchedTemplate](https://github.com/mohammaduzair9/Basic-Digital-Image-Processing/blob/master/Template%20Matching/matchedTemplate.png)|\n\n" }, { "path": "Sharpening/sharpen.py", "content": "import cv2\nimport numpy as np\nfrom matplotlib import pyplot as plt\nimport copy\n\nplt.figure(figsize=(12,12))\n#reading image file\nim = cv2.imread(\"inp1.jpg\", 1)\n#function for sharpening filter\ndef sharpenFiltering(img):\n inputImg = copy.deepcopy(img.astype(np.float))\n #converting color scale from BGR to GRAY\n inputImg = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)\n #initialize black image of size equal to given image\n outputImg = np.zeros(inputImg.shape)\n #padding the image with zeros\n inputImg = np.pad(inputImg, (1, 1), 'constant', constant_values=(0))\n #creating two filters for horizontal and vertical edge detection\n fh = np.array([[-1.0,-2.0,-1.0],[0.0,0.0,0.0],[1.0,2.0,1.0]])\n fy = np.array([[-1.0,0.0,1.0],[-2.0,0.0,2.0],[-1.0,0.0,1.0]])\n #looping through image pixels\n for row in range(1, inputImg.shape[0]-1):\n for col in range(1, inputImg.shape[1]-1):\n dx, dy = 0.0, 0.0\n #convolving both filters\n for x_filter in xrange(3):\n for y_filter in xrange(3):\n dx += inputImg[row+x_filter-1][col+y_filter-1]*fh[x_filter][y_filter]\n dy += inputImg[row+x_filter-1][col+y_filter-1]*fy[x_filter][y_filter]\n \n #magnitude of gradient (instead of just adding dx and dy. we calculate magnitude)\n pixel = np.sqrt(dx * dx + dy * dy)\n outputImg[row-1][col-1] = pixel\n #normalizing pixels\n outputImg *= 255.0/np.max(outputImg)\n return outputImg\n\n#applying sharpen filters\noutput = sharpenFiltering(im)\n#writing image to image file\ncv2.imwrite(\"sharpen.jpg\",output)\n#converting color scale from BGR to RGB\nim = cv2.cvtColor(im, cv2.COLOR_BGR2RGB)\n#plotting original image\nplt.subplot(211)\nplt.axis('off')\nplt.title(\"Original Image\")\nplt.imshow(im)\n#plotting transformed image\nplt.subplot(212)\nplt.axis('off')\nplt.title(\"Transformed Image\")\nplt.imshow(output, cmap=\"gray\")\n\nplt.show()" }, { "path": "Skeletonization/Skeletonization.py", "content": "import numpy as np\nimport cv2\n\nimg = cv2.imread('Thumb.png',0)\nret,img = cv2.threshold(img,127,255,cv2.THRESH_BINARY)\n\n# Function for skeletonizing the image\ndef findSkeleton(im):\n element = cv2.getStructuringElement(cv2.MORPH_CROSS,(3,3))\n out = np.zeros(im.shape,np.uint8)\n\n flag = 0 \n while(not flag):\n eroded = cv2.erode(im, element)\n opened = cv2.dilate(eroded, element)\n opened = cv2.subtract(im,opened)\n out = cv2.bitwise_or(out,opened)\n im = eroded.copy()\n zeros = img.size - cv2.countNonZero(im)\n flag = 1 if (zeros == img.size) else 0\n return out\n\noutput = findSkeleton(img)\n\nkernel = np.ones((3,3),np.uint8)\noutput = cv2.dilate(output,kernel)\noutput = cv2.medianBlur(output, 5)\nret,thresh = cv2.threshold(output,127,255,cv2.THRESH_BINARY_INV)\n\nres = np.hstack((img, thresh))\ncv2.imshow(\"output\", cv2.resize(res, dsize=None,fx=0.5, fy=0.5))\ncv2.imwrite(\"task1_output.png\", res)\ncv2.waitKey(0)\n\n" }, { "path": "Smoothing/AvergingFilter.py", "content": "import cv2\nimport numpy as np\nfrom matplotlib import pyplot as plt\nfrom Filter import applyFilter\n\nplt.figure(figsize=(12,12))\n\n#reading image from file\nim = cv2.imread(\"inp1.tif\", 0).astype(np.float)\n\nsize = int(raw_input(\"> Enter the size of averaging filter: \"))\n#applying filter on image\noutput = applyFilter(im, filterSize=size)\n\n#writing image to image file\ncv2.imwrite(\"averaging.jpg\",output)\n\n#plotting original image\nplt.subplot(211)\nplt.axis('off')\nplt.title(\"Original Image\")\nplt.imshow(im, cmap=\"gray\")\n\n#plotting smoothed image\nplt.subplot(212)\nplt.axis('off')\nplt.title(\"Smoothed Image (avg. filter\"+str(size)+\"x\"+str(size)+\")\")\nplt.imshow(output, cmap=\"gray\")\n\nplt.show()" }, { "path": "Smoothing/Filter.py", "content": "import numpy as np\nimport copy\n\n#function to apply filter on image\n#call it with just filter size (averaging filter)\n#call it with given filter in imFilter\ndef applyFilter(img, filterSize=None, imFilter=None):\n filteredImg = copy.deepcopy(img)\n #if filter is provided\n if (imFilter is not None):\n imgFilter = imFilter\n filterSize = len(imFilter)\n #if filter is not provided, create an averging filter\n else:\n imgFilter = (1.0/(filterSize*filterSize))*np.ones((filterSize,filterSize)) #creating filter\n \n paddingSize = filterSize/2\n #padding the image with zeros\n filteredImg = np.pad(filteredImg, (paddingSize, paddingSize), 'constant', constant_values=(0))\n \n for row in range(paddingSize, filteredImg.shape[0]-paddingSize):\n for col in range(paddingSize, filteredImg.shape[1]-paddingSize):\n pixel = 0.0\n #convolving the filter\n for x_filter in xrange(filterSize):\n for y_filter in xrange(filterSize):\n pixel += filteredImg[row+x_filter-paddingSize][col+y_filter-paddingSize]*imgFilter[x_filter][y_filter]\n filteredImg[row,col] = pixel\n #removing padded pixels\n filteredImg = filteredImg[paddingSize:filteredImg.shape[0]-paddingSize, paddingSize:filteredImg.shape[1]-paddingSize]\n return filteredImg" }, { "path": "Smoothing/gaussian.py", "content": "import cv2\nimport numpy as np\nfrom matplotlib import pyplot as plt\nfrom Filter import applyFilter\n\nplt.figure(figsize=(12,12))\n\nim = cv2.imread(\"inp1.tif\", 0).astype(np.float)\n\n#creating gaussian filter \ngaussianFilter = np.array([[1,1,2,2,2,1,1],\n [1,2,2,4,2,2,1],\n [2,2,4,8,4,2,2],\n [2,4,8,16,8,4,2],\n [2,2,4,8,4,2,2],\n [1,2,2,4,2,2,1],\n [1,1,2,2,2,1,1]], np.float)\ngaussianFilter /= np.sum(gaussianFilter*1.0)\n\n#applying gaussian filter\noutput = applyFilter(im, imFilter=gaussianFilter)\n#writing image to image file\ncv2.imwrite(\"gaussian.jpg\",output)\n#plotting original image\nplt.subplot(211)\nplt.axis('off')\nplt.title(\"Original Image\")\nplt.imshow(im, cmap=\"gray\")\n\n#plotting smoothed image \nplt.subplot(212)\nplt.axis('off')\nplt.title(\"Smoothed Image (gaussian filter7x7 (sigma 1.4))\")\nplt.imshow(output, cmap=\"gray\")\nplt.show()" }, { "path": "Smoothing/median.py", "content": "import cv2\nimport numpy as np\nfrom matplotlib import pyplot as plt\nimport copy\n\nplt.figure(figsize=(12,12))\n#reading image from file\nim = cv2.imread(\"inp3.tif\", 0).astype(np.float)\n#function for median filtering\ndef medianFiltering(img, filterSize):\n #making deep copy of image\n filteredImg = copy.deepcopy(img)\n #calculating padding size\n paddingSize = filterSize/2\n #padding the image\n filteredImg = np.pad(filteredImg, (paddingSize, paddingSize), 'constant', constant_values=(0))\n #loop through image pixels\n for row in range(paddingSize, filteredImg.shape[0]-paddingSize):\n for col in range(paddingSize, filteredImg.shape[1]-paddingSize):\n kernal = []\n for x_filter in xrange(filterSize):\n for y_filter in xrange(filterSize):\n kernal.append(filteredImg[row+x_filter-paddingSize][col+y_filter-paddingSize])\n #calculating median of list\n filteredImg[row,col] = np.median(kernal)\n #removing zero padding \n filteredImg = filteredImg[paddingSize:filteredImg.shape[0]-paddingSize, paddingSize:filteredImg.shape[1]-paddingSize]\n return filteredImg\n\nsize = int(raw_input(\"> Enter the size of median filter: \"))\n#applying meadian filtering on image\noutput = medianFiltering(im, filterSize=size)\n#writing file to image file\ncv2.imwrite(\"median.jpg\",output)\n#plotting original image\nplt.subplot(211)\nplt.axis('off')\nplt.title(\"Original Image\")\nplt.imshow(im, cmap=\"gray\")\n#plotting transformed image\nplt.subplot(212)\nplt.axis('off')\nplt.title(\"Transformed Image\")\nplt.imshow(output, cmap=\"gray\")\n\nplt.show()" }, { "path": "Smoothing/unsharp_masking.py", "content": "import cv2\nimport numpy as np\nfrom matplotlib import pyplot as plt\nimport copy\nfrom Filter import applyFilter\n\nplt.figure(figsize=(12,12))\n#reading image from file\nim = cv2.imread(\"inp2.tif\", 0).astype(np.float)\n\n#function for unsharp masking\ndef unsharpMasking(img):\n inputImg = copy.deepcopy(img)\n blurredImg = applyFilter(inputImg, filterSize=5)\n mask = inputImg - blurredImg\n result = inputImg + mask\n return result\n#unsharp masking the image\noutput = unsharpMasking(im)\n#writing image to image file\ncv2.imwrite(\"unsharp_masking.jpg\",output)\n#plotting original image\nplt.subplot(211)\nplt.axis('off')\nplt.title(\"Original Image\")\nplt.imshow(im, cmap=\"gray\")\n#plotting transformed image\nplt.subplot(212)\nplt.axis('off')\nplt.title(\"Transformed Image\")\nplt.imshow(output, cmap=\"gray\")\n\nplt.show()" }, { "path": "Smoothing/weightedavg.py", "content": "import cv2\nimport numpy as np\nfrom matplotlib import pyplot as plt\nfrom Filter import applyFilter\n\nplt.figure(figsize=(12,12))\n\n#reading image from file\nim = cv2.imread(\"inp1.tif\", 0).astype(np.float)\n\n#creating weighted filter:\n#[1 2 1],\n#[2 4 2],\n#[1 2 1]\nweightedFilter = (1.0/16)*np.array([[1,2,1],[2,4,2],[1,2,1]], np.int32)\n\n#applying filter on image\noutput = applyFilter(im, imFilter=weightedFilter)\n\n#writing image to image file\ncv2.imwrite(\"weightedavg.jpg\",output)\n\n#plotting original image\nplt.subplot(211)\nplt.axis('off')\nplt.title(\"Original Image\")\nplt.imshow(im, cmap=\"gray\")\n\n#plotting smoothed image\nplt.subplot(212)\nplt.axis('off')\nplt.title(\"Smoothed Image (weighted avg. filter3x3)\")\nplt.imshow(output, cmap=\"gray\")\nplt.show()" }, { "path": "Template Matching/TemplateMatching.py", "content": "import cv2\nimport numpy as np\n\nimage = cv2.imread('image.png')\ntemplate = cv2.imread('template.png')\n(templateHeight, templateWidth) = template.shape[:2]\n\nmatchResult = cv2.matchTemplate(image, template, cv2.TM_CCOEFF)\n(_, _, minLoc, maxLoc) = cv2.minMaxLoc(matchResult)\n\ntopLeft = maxLoc\nbotRight = (topLeft[0] + templateWidth, topLeft[1] + templateHeight)\nroi = image[topLeft[1]:botRight[1], topLeft[0]:botRight[0]]\n \nmask = np.zeros(image.shape, dtype = \"uint8\")\nimage = cv2.addWeighted(image, 0.25, mask, 0.75, 0)\n\nimage[topLeft[1]:botRight[1], topLeft[0]:botRight[0]] = roi\n \ncv2.imwrite(\"matchedTemplate.png\", image)\n" }, { "path": "XY_Cuts/XY_Cuts.py", "content": "import cv2\nfrom matplotlib import pyplot as plt\n\ndef xycut(image, image_path):\n # reading image\n img = plt.imread(image_path)\n fig, ax = plt.subplots()\n ax.imshow(img)\n\n black_pix = []\n white_lines = []\n # detecting the white lines\n for i in range(0, bin_img.shape[0]):\n # if number of white phixels in row is greater than 750\n if (bin_img[i].sum() / 255 > 750):\n\n # draw horizontal lines\n ax.axhline(y=i, color='green')\n else:\n # black pixels on x-axis\n for j in range(0, bin_img.shape[1]):\n if (bin_img[i][j] == 0):\n black_pix.append(j)\n\n if len(black_pix) != 0:\n # draw first & last vertical line only\n ax.axvline(x=min(black_pix), linewidth=3, color='green')\n ax.axvline(x=max(black_pix), linewidth=3, color='green')\n\n # remove the axis\n ax.set_axis_off()\n # saving new figure\n plt.savefig('xycut.png', bbox_inches='tight')\n # show the figure\n plt.show()\n\n\n# load image as greyscale\nimage = cv2.imread(\"XY-cuts.png\", 0)\n# binarize image\n(_, bin_img) = cv2.threshold(image, 120, 255, cv2.THRESH_BINARY)\n# call xycut function\nxycut(bin_img, \"XY-cuts.png\")\n" } ]