OpenCV簡單標準數字識別的完整實例

更新時間：2021年09月09日 11:40:51 作者：huang_nansen

這篇文章主要給大家介紹了關于OpenCV簡單標準數字識別的相關資料,要通過opencv 進行數字識別離不開訓練庫的支持，需要對目標圖片進行大量的訓練，才能做到精準的識別出目標數字,需要的朋友可以參考下

在學習openCV時，看到一個問答做數字識別，里面配有代碼，應用到了openCV里面的ml包，很有學習價值。

https://stackoverflow.com/questions/9413216/simple-digit-recognition-ocr-in-opencv-python#

import sys
import numpy as np
import cv2
 
im = cv2.imread('t.png')
im3 = im.copy()
 
gray = cv2.cvtColor(im,cv2.COLOR_BGR2GRAY)   #先轉換為灰度圖才能夠使用圖像閾值化
 
thresh = cv2.adaptiveThreshold(gray,255,cv2.ADAPTIVE_THRESH_GAUSSIAN_C,cv2.THRESH_BINARY,11,2)  #自適應閾值化
 
##################      Now finding Contours         ###################
# 
image,contours,hierarchy = cv2.findContours(thresh,cv2.RETR_LIST,cv2.CHAIN_APPROX_SIMPLE)
#邊緣查找，找到數字框，但存在誤判
 
samples =  np.empty((0,900))    #將每一個識別到的數字所有像素點作為特征，儲存到一個30*30的矩陣內
responses = []                  #label
keys = [i for i in range(48,58)]    #48-58為ASCII碼
count =0
for cnt in contours:
    if cv2.contourArea(cnt)>80:     #使用邊緣面積過濾較小邊緣框
        [x,y,w,h] = cv2.boundingRect(cnt)   
        if  h>25 and h < 30:        #使用高過濾小框和大框
            count+=1
            cv2.rectangle(im,(x,y),(x+w,y+h),(0,0,255),2)
            roi = thresh[y:y+h,x:x+w]
            roismall = cv2.resize(roi,(30,30))
            cv2.imshow('norm',im)
            key = cv2.waitKey(0)
            if key == 27:  # (escape to quit)
                sys.exit()
            elif key in keys:
                responses.append(int(chr(key)))
                sample = roismall.reshape((1,900))
                samples = np.append(samples,sample,0)
            if count == 100:        #過濾一下過多邊緣框，后期可能會嘗試極大抑制
                break
responses = np.array(responses,np.float32)
responses = responses.reshape((responses.size,1))
print ("training complete")
 
np.savetxt('generalsamples.data',samples)
np.savetxt('generalresponses.data',responses)
#
cv2.waitKey()
cv2.destroyAllWindows()

訓練數據為：

測試數據為：

使用openCV自帶的ML包，KNearest算法

 
import sys
import cv2
import numpy as np
 #######   training part    ############### 
samples = np.loadtxt('generalsamples.data',np.float32)
responses = np.loadtxt('generalresponses.data',np.float32)
responses = responses.reshape((responses.size,1))
 
model = cv2.ml.KNearest_create()
model.train(samples,cv2.ml.ROW_SAMPLE,responses)
 
 
def getNum(path):
    im = cv2.imread(path)
    out = np.zeros(im.shape,np.uint8)
    gray = cv2.cvtColor(im,cv2.COLOR_BGR2GRAY)
    
    #預處理一下
    for i in range(gray.__len__()):
        for j in range(gray[0].__len__()):
            if gray[i][j] == 0:
                gray[i][j] == 255
            else:
                gray[i][j] == 0
    thresh = cv2.adaptiveThreshold(gray,255,1,1,11,2)
     
    image,contours,hierarchy = cv2.findContours(thresh,cv2.RETR_LIST,cv2.CHAIN_APPROX_SIMPLE)
    count = 0 
    numbers = []
    for cnt in contours:
        if cv2.contourArea(cnt)>80:
            [x,y,w,h] = cv2.boundingRect(cnt)
            if  h>25:
                cv2.rectangle(im,(x,y),(x+w,y+h),(0,255,0),2)
                roi = thresh[y:y+h,x:x+w]
                roismall = cv2.resize(roi,(30,30))
                roismall = roismall.reshape((1,900))
                roismall = np.float32(roismall)
                retval, results, neigh_resp, dists = model.findNearest(roismall, k = 1)
                string = str(int((results[0][0])))
                numbers.append(int((results[0][0])))
                cv2.putText(out,string,(x,y+h),0,1,(0,255,0))
                count += 1
        if count == 10:
            break
    return numbers
 
numbers = getNum('1.png')