Python OpenCV Answer Sheet Recognition: Preprocessing, Contour Detection, Perspective Transform, and Option Detection

The article demonstrates a step‑by‑step method for automatically reading answer sheet images using Python and OpenCV, covering image preprocessing, contour detection, perspective correction, and option detection to produce a 5×5 answer matrix.

1. Preprocessing (denoise, grayscale, binarization)

img = cv2.imread("1.png",1)
#高斯去噪
img_gs = cv2.GaussianBlur(img,[5,5],0)
# 转灰度
img_gray = cv2.cvtColor(img_gs,cv2.COLOR_BGR2GRAY)
# 自适应二值化
_,binary_img = cv2.threshold(img_gray,0,255,cv2.THRESH_OTSU|cv2.THRESH_BINARY)

Note: cv2.THRESH_OTSU|cv2.THRESH_BINARY applies Otsu's adaptive thresholding with inverse binary output.

2. Contour Detection

#找轮廓
contours, hierarchy = cv2.findContours(binary_img,cv2.RETR_EXTERNAL,cv2.CHAIN_APPROX_NONE)
#按照轮廓的面积从大到小排序
cnts = sorted(contours,key = cv2.contourArea,reverse=True)
#画轮廓
draw_img = cv2.drawContours(img.copy(),cnts[0],-1,(0,255,255),2)

The findContours function expects a binary image; cv2.RETR_EXTERNAL extracts only external contours, and cv2.CHAIN_APPROX_NONE returns all contour points.

3. Perspective Transform

#透视变换
#矩形的四个顶点为approxCurve[0][0],approxCurve[1][0],approxCurve[2][0],approxCurve[3][0]
#分别表示矩形的TL,BL,BR,TR四个点
a1 = list(approxCurve[0][0])
a2 = list(approxCurve[1][0])
a3 = list(approxCurve[2][0])
a4 = list(approxCurve[3][0])
#原始矩阵
mat1 = np.array([a1,a2,a3,a4],dtype = np.float32)
#计算矩形的w和h
w1 = int(np.sqrt((a1[0]-a4[0])**2+(a1[1]-a4[1])**2))
w2 = int(np.sqrt((a2[0]-a3[0])**2+(a2[1]-a3[1])**2))
h1 = int(np.sqrt((a1[0]-a2[0])**2+(a1[1]-a2[1])**2))
h2 = int(np.sqrt((a3[0]-a4[0])**2+(a3[1]-a4[1])**2))
w,h = max(w1,w2),max(h1,h2)
#目标矩阵
new_a1 = [0,0]
new_a2 = [0,h]
new_a3 = [w,h]
new_a4 = [w,0]
mat2 = np.array([new_a1,new_a2,new_a3,new_a4],dtype = np.float32)
#透视变换矩阵
mat = cv2.getPerspectiveTransform(mat1,mat2)
#进行透视变换
res = cv2.warpPerspective(img,mat,(w,h))
imshow((res))

The transformation aligns the answer sheet, removing surrounding background and correcting perspective distortion.

4. Detecting Each Option Contour

res_gray = cv2.cvtColor(res,cv2.COLOR_BGR2GRAY)
_,binary_res = cv2.threshold(res_gray,0,255,cv2.THRESH_OTSU|cv2.THRESH_BINARY_INV)
contours = cv2.findContours(binary_res,cv2.RETR_EXTERNAL,cv2.CHAIN_APPROX_NONE)[0]
#挑选合适的轮廓
def check(contours):
    ans = []
    for i in contours:
        area = float(cv2.contourArea(i))
        length = float(cv2.arcLength(i,True))
        if area<=0 or length<=0:
            continue
        if area/length > 7.05 and area/length < 10.5:
            ans.append(i)
    return ans
ans_contours = check(contours)

5. Drawing Enclosing Circles, Sorting, and Locating Options

#遍历每一个圆形轮廓，画外接圆
circle = []
for i in ans_contours:
    (x,y),r = cv2.minEnclosingCircle(i)
    center = (int(x),int(y))
    r = int(r)
    circle.append((center,r))
#按照外接圆的水平坐标排序center[1]（y坐标）
circle.sort(key = lambda x:x[0][1])
A = []
for i in range(1,6):
    now = circle[(i-1)*5:i*5]
    now.sort(key = lambda x:x[0][0])
    A.extend(now)

The sorted list A now holds the option circles from left‑to‑right and top‑to‑bottom.

6. Option Detection

def dots_distance(dot1,dot2):
    #计算二维空间中两个点的距离
    return ((dot1[0]-dot2[0])**2+(dot1[1]-dot2[1])**2)**0.5

def count_dots(center,radius):
    #输入圆的中心点与半径，返回圆内所有的坐标
    dots = []
    for i in range(-radius,radius+1):
        for j in range(-radius,radius+1):
            dot2 = (center[0]+i,center[1]+j)
            if dots_distance(center,dot2) <= radius:
                dots.append(dot2)
    return dots

da = []
for i in A:
    dots = count_dots(i[0],i[1])
    all_dots = len(dots)
    white_dots = 0
    for j in dots:
        if binary_res[j[1]][j[0]] == 255:
            white_dots += 1
    if white_dots/all_dots >= 0.4:
        da.append(1)
    else:
        da.append(0)

da = np.array(da)
da = np.reshape(da,(5,5))

The resulting 5×5 NumPy array da represents the selected answers, where 1 indicates a marked option and 0 indicates an unmarked one.