利用PIL做简单的图片数字识别

#coding:gbk
'''
Created on 2009-7-7

@author: yuhai
'''
import StringIO
import Image,os
import editdist

dic={}


def readimg(imgpath):
        buffer=StringIO.StringIO();
        im = Image.open(imgpath)
        for y in range(20):
            for x in range(9):
                v= im.getpixel((x,y))
                if v<200:buffer.write('1')
                else : buffer.write('0')
        buffer.seek(0)
        return buffer.read()

def regimg(imgpath):
    im = Image.open(imgpath).convert('L')
    result=StringIO.StringIO()
    for i in range(4):
        tmp =im.crop((i*9,0,i*10+10,20))
        s=StringIO.StringIO()
        tmp.save(s,"JPEG")
        s.seek(0)
        v=readimg(s)
        idx=0      
        score=editdist.distance(v,dic[0])
        for j in range(1,10):
            tmp=editdist.distance(v,dic[j])
            if tmp<score:
                score=tmp
                idx=j
        result.write(str(idx))
    result.seek(0)
    return result.read()
def init():
    f=open('./data','w')
    for i in range(10):
        dic[i]= readimg('e:/'+str(i)+'.bmp')
        f.write(dic[i])
        f.write('\n')
    f.close()

def readdata():
    f=open('./data','r')
    i=0
    for line in f.readlines():
        dic[i]= line.strip()
        i=i+1
    f.close()

def test():
    files = os.listdir('e:/img/') 
    redic=[]
    for f in files: 
        if f.find('.bmp') !=-1:
            result=regimg('e:/img/'+f)
            if f==result+'.bmp':
                redic.append(f)
                print f,result
    print 'correct %s' % len(redic)

def getcontent(imgpath):
    im = Image.open(imgpath).convert('L')
    print im.size
    box=(10,0,19,20)
    t=im.crop(box)
    t.save('e:/5.bmp', "JPEG") 

if __name__ == '__main__':   
    #getcontent('e:/img/1560.bmp')
    #init()
    readdata()
    test()