The practice of OpenCV -- bank card number recognition

First, let's introduce the library we need and the image display we define （ convenient ）：

import cv2
import numpy as np
import imutils
from imutils import contours

# Display images 
def cv_show(img,name="image"):
    cv2.imshow(name,img)
    cv2.waitKey(0)
    cv2.destroyWindow(name)

  Secondly, let's introduce the template of bank card numbers and carry out preprocessing ：

 

template=cv2.imread('TT.jpg')
# cv_show(template)
# Convert to grayscale 
ref_template=cv2.cvtColor(template,cv2.COLOR_BGR2GRAY)
# cv_show(ref_template)
# Convert to binary graph 
ref_template=cv2.threshold(ref_template,10,255,cv2.THRESH_BINARY_INV)[1]
# cv_show(ref_template)

# Contour detection 
##findContours  Only accept binary graphs    RETR_EXTERNAL  Only the outer contour is detected    CHAIN_APPROX_SIMPLE  Keep only the end coordinates 
BIT,refCnts,hierarchy=cv2.findContours(ref_template.copy(),cv2.RETR_EXTERNAL,cv2.CHAIN_APPROX_SIMPLE)
cv2.drawContours(template,refCnts,-1,(0,0,255),3)
# cv_show(template)
# print(np.array(refCnts).shape)
# Sort outline 
refCnts=contours.sort_contours(refCnts,method="left-to-right")[0]
# print(refCnts)

digits={}
for (i,c) in enumerate(refCnts):
    # Calculate the circumscribed rectangle 
    (x,y,w,h)=cv2.boundingRect(c)
    roi=ref_template[y:y+h,x:x+w]
    #resize To the right size 
    roi=cv2.resize(roi,(57,88))
     # cv_show(roi)
    # Each number corresponds to a template 
    digits[i]=roi

  Next, introduce and process the bank card we need to identify ：

# Initialize convolution kernel 
rectKernel=cv2.getStructuringElement(cv2.MORPH_RECT,(9,3))
sqKernel=cv2.getStructuringElement(cv2.MORPH_RECT,(5,5))

# Read input image 
image=cv2.imread('T1.jpg')

# cv_show(image)
image=imutils.resize(image,width=300)
image_gray=cv2.cvtColor(image,cv2.COLOR_BGR2GRAY)
# cv_show(image_gray)

#tophat Top hat operation , Highlight bright areas 
image_tophat=cv2.morphologyEx(image_gray,cv2.MORPH_TOPHAT,rectKernel)
# cv_show(image_tophat)

# Calculate the gradient    Make the outline clearer 
gradX=cv2.Sobel(image_tophat,ddepth=cv2.CV_32F,dx=1,dy=0,ksize=-1)#ksize=-1 To express with 3*3 And nucleus 
gradX=np.absolute(gradX)
(minVal,maxVal)=(np.min(gradX),np.max(gradX))
gradX=(255*(gradX-minVal)/(maxVal-minVal))
gradX=gradX.astype("uint8")
# print(np.array(gradX).shape)
# cv_show(gradX)

# Close your operations （ Expand first and then corrode ） Put the numbers together 
gradX=cv2.morphologyEx(gradX,cv2.MORPH_CLOSE,rectKernel)
# cv_show(gradX)

# Two valued 
image_thresh=cv2.threshold(gradX,0,255,cv2.THRESH_BINARY|cv2.THRESH_OTSU)[1]
# cv_show(image_thresh)

# Close again after binarization 
image_thresh=cv2.morphologyEx(image_thresh,cv2.MORPH_CLOSE,sqKernel)
# cv_show(image_thresh)

# Calculate the contour 
BITT,threshCnts,hierarchy=cv2.findContours(image_thresh.copy(),cv2.RETR_EXTERNAL,cv2.CHAIN_APPROX_SIMPLE)


cv2.drawContours(image.copy(),threshCnts,-1,(0,0,255),1)
# cv_show(image)

  Finally, let's match the bank card template and the digital template ：

locs=[]
for (i,c) in enumerate(threshCnts):
    (x,y,w,h)=cv2.boundingRect(c)
    ar=w/float(h)
    if ar>2.5 and ar<4.0:
        if (w>40 and w<55 ) and (h>10 and h<20):
            locs.append((x,y,w,h))
locs=sorted(locs,key=lambda x:x[0])
output=[]

for (i,(gx,gy,gw,gh)) in enumerate(locs):# gx,gy,gw,gh  representative  4000  This area 
    groupOutput=[]
    # Extract each set of numbers （ One group includes 4 A digital ）
    group=image_gray[gy-5:gy+gh+5,gx-5:gx+gw+5]
    # cv_show(group)

    group=cv2.threshold(group,0,255,cv2.THRESH_BINARY|cv2.THRESH_OTSU)[1]
    # cv_show(group)

    B,digitsCnts,A=cv2.findContours(group.copy(),cv2.RETR_EXTERNAL,cv2.CHAIN_APPROX_SIMPLE)
    digitsCnts=imutils.contours.sort_contours(digitsCnts,method="left-to-right")[0]
    # print(digitsCnts)
    for c in digitsCnts:  # Go through every number     Such as 4000   Traverse in turn 4、0、0、0
        (x,y,w,h)=cv2.boundingRect(c)
        # print((x,y,w,h))
        roi=group[y:y+h,x:x+w]
        # Set the matching image to the consistent size of the template image 
        roi=cv2.resize(roi,(57,88))
        # cv_show(roi)

        # Calculate the match score 
        scores=[]

        for (digit,digitROI) in digits.items():
            # Template matching 
            result=cv2.matchTemplate(roi,digitROI,cv2.TM_CCOEFF)
            (_,score,_,_)=cv2.minMaxLoc(result)
            scores.append(score)

        groupOutput.append(str(np.argmax(scores)))

    cv2.rectangle(image,(gx-5,gy-5),(gx+gw+5,gy+gh+5),(0,0,255),1)
    cv2.putText(image,"".join(groupOutput),(gx,gy-15),cv2.FONT_HERSHEY_SIMPLEX,0.65,(0,0,255),2)

    output.extend(groupOutput)

cv_show(image)
print("".join(output))

 

  The result shown at last .

Full code ：

import cv2
import numpy as np
import imutils
from imutils import contours

# Display images 
def cv_show(img,name="image"):
    cv2.imshow(name,img)
    cv2.waitKey(0)
    cv2.destroyWindow(name)


template=cv2.imread('TT.jpg')
# cv_show(template)
# Convert to grayscale 
ref_template=cv2.cvtColor(template,cv2.COLOR_BGR2GRAY)
# cv_show(ref_template)
# Convert to binary graph 
ref_template=cv2.threshold(ref_template,10,255,cv2.THRESH_BINARY_INV)[1]
# cv_show(ref_template)

# Contour detection 
##findContours  Only accept binary graphs    RETR_EXTERNAL  Only the outer contour is detected    CHAIN_APPROX_SIMPLE  Keep only the end coordinates 
BIT,refCnts,hierarchy=cv2.findContours(ref_template.copy(),cv2.RETR_EXTERNAL,cv2.CHAIN_APPROX_SIMPLE)
cv2.drawContours(template,refCnts,-1,(0,0,255),3)
# cv_show(template)
# print(np.array(refCnts).shape)
# Sort outline 
refCnts=contours.sort_contours(refCnts,method="left-to-right")[0]
# print(refCnts)

digits={}
for (i,c) in enumerate(refCnts):
    # Calculate the circumscribed rectangle 
    (x,y,w,h)=cv2.boundingRect(c)
    roi=ref_template[y:y+h,x:x+w]
    #resize To the right size 
    roi=cv2.resize(roi,(57,88))
     # cv_show(roi)
    # Each number corresponds to a template 
    digits[i]=roi

# Initialize convolution kernel 
rectKernel=cv2.getStructuringElement(cv2.MORPH_RECT,(9,3))
sqKernel=cv2.getStructuringElement(cv2.MORPH_RECT,(5,5))

# Read input image 
image=cv2.imread('T1.jpg')

# cv_show(image)
image=imutils.resize(image,width=300)
image_gray=cv2.cvtColor(image,cv2.COLOR_BGR2GRAY)
# cv_show(image_gray)

#tophat Top hat operation , Highlight bright areas 
image_tophat=cv2.morphologyEx(image_gray,cv2.MORPH_TOPHAT,rectKernel)
# cv_show(image_tophat)

# Calculate the gradient    Make the outline clearer 
gradX=cv2.Sobel(image_tophat,ddepth=cv2.CV_32F,dx=1,dy=0,ksize=-1)#ksize=-1 To express with 3*3 And nucleus 
gradX=np.absolute(gradX)
(minVal,maxVal)=(np.min(gradX),np.max(gradX))
gradX=(255*(gradX-minVal)/(maxVal-minVal))
gradX=gradX.astype("uint8")
# print(np.array(gradX).shape)
# cv_show(gradX)

# Close your operations （ Expand first and then corrode ） Put the numbers together 
gradX=cv2.morphologyEx(gradX,cv2.MORPH_CLOSE,rectKernel)
# cv_show(gradX)

# Two valued 
image_thresh=cv2.threshold(gradX,0,255,cv2.THRESH_BINARY|cv2.THRESH_OTSU)[1]
# cv_show(image_thresh)

# Close again after binarization 
image_thresh=cv2.morphologyEx(image_thresh,cv2.MORPH_CLOSE,sqKernel)
# cv_show(image_thresh)

# Calculate the contour 
BITT,threshCnts,hierarchy=cv2.findContours(image_thresh.copy(),cv2.RETR_EXTERNAL,cv2.CHAIN_APPROX_SIMPLE)


cv2.drawContours(image.copy(),threshCnts,-1,(0,0,255),1)
# cv_show(image)


locs=[]
for (i,c) in enumerate(threshCnts):
    (x,y,w,h)=cv2.boundingRect(c)
    ar=w/float(h)
    if ar>2.5 and ar<4.0:
        if (w>40 and w<55 ) and (h>10 and h<20):
            locs.append((x,y,w,h))
locs=sorted(locs,key=lambda x:x[0])
output=[]

for (i,(gx,gy,gw,gh)) in enumerate(locs):# gx,gy,gw,gh  representative  4000  This area 
    groupOutput=[]
    # Extract each set of numbers （ One group includes 4 A digital ）
    group=image_gray[gy-5:gy+gh+5,gx-5:gx+gw+5]
    # cv_show(group)

    group=cv2.threshold(group,0,255,cv2.THRESH_BINARY|cv2.THRESH_OTSU)[1]
    # cv_show(group)

    B,digitsCnts,A=cv2.findContours(group.copy(),cv2.RETR_EXTERNAL,cv2.CHAIN_APPROX_SIMPLE)
    digitsCnts=imutils.contours.sort_contours(digitsCnts,method="left-to-right")[0]
    # print(digitsCnts)
    for c in digitsCnts:  # Go through every number     Such as 4000   Traverse in turn 4、0、0、0
        (x,y,w,h)=cv2.boundingRect(c)
        # print((x,y,w,h))
        roi=group[y:y+h,x:x+w]
        # Set the matching image to the consistent size of the template image 
        roi=cv2.resize(roi,(57,88))
        # cv_show(roi)

        # Calculate the match score 
        scores=[]

        for (digit,digitROI) in digits.items():
            # Template matching 
            result=cv2.matchTemplate(roi,digitROI,cv2.TM_CCOEFF)
            (_,score,_,_)=cv2.minMaxLoc(result)
            scores.append(score)

        groupOutput.append(str(np.argmax(scores)))

    cv2.rectangle(image,(gx-5,gy-5),(gx+gw+5,gy+gh+5),(0,0,255),1)
    cv2.putText(image,"".join(groupOutput),(gx,gy-15),cv2.FONT_HERSHEY_SIMPLEX,0.65,(0,0,255),2)

    output.extend(groupOutput)

cv_show(image)
print("".join(output))

  Different versions findContours Functions may be used in different ways , Some higher versions require only two return values .

All the above are used in the past opencv What I learned , There may also be something you haven't learned ,