scbc_repos/ssat_sdk/picture/ocr_tesseract.py

# -*- coding:utf-8 -*-
from ssat_sdk import sat_environment
from ssat_sdk.utils.LoggingUtil import DebugLogger

import os, sys, time
import pytesseract as tes
import cv2 as cv
try:
    from PIL import Image
except ImportError:
    import Image

Lan_Default = ""
Tes_LanDIC = {"chineseprc+english": "chi_eng", "chinesetaiwan+english": "",
                "spanish": "",
                "chineseprc": "chi_sim", "chinesetaiwan": "chi_tra", "russian": "", "french": "",
                "english": "eng", "vietnamese": "", "hebrew": "", "thai": "",
                "arabic":""
                }

class OCRTes():
    def __init__(self):
        self.logger = DebugLogger()
        self.uiStyle = sat_environment.getMenuTreeSelectedStyle()
        self.chip = sat_environment.getMenuTreeSelectedChip()
        self.tesData_proj = self.chip + "." + self.uiStyle
        self.logger.info(__file__, "OCRTes", "__init__", "tesData:" + self.tesData_proj)
        # print "getMenuTreeRootDir:",sat_environment.getMenuTreeRootDir()
        # print "getMenuTreeSelectedProjectCfgPath:",sat_environment.getMenuTreeSelectedProjectCfgPath()
        # print "getMenuTreeSelectedChip:",sat_environment.getMenuTreeSelectedChip()
        # print "getMenuTreeSelectedTVUI:",sat_environment.getMenuTreeSelectedTVUI()
        # print "getSATMenuTreeDIR:",sat_environment.getSATMenuTreeDIR()


    def getStr(self,picPath, lan, type ):
        tesdata = self.tesData_proj #+ "." + Lan_Default #每个项目，暂时只放一个字库，不区分语言
        print "OCRTes.getStr,params:",picPath, tesdata, type
        startTime = time.time()
        strArr = tes.image_to_string(picPath, lang=tesdata)
        print "OCRTes,getStr cost:", time.time() - startTime
        # strArr = tes.image_to_data(img, lang=lan)
        # print strArr
        return strArr