OCR实现微信截图改名

bash 复制代码
pip install paddlepaddle -i https://pypi.tuna.tsinghua.edu.cn/simple/                                                                                   ──(Sat,Nov30)─┘
pip install shapely -i https://pypi.tuna.tsinghua.edu.cn/simple/
pip install paddleocr -i https://pypi.tuna.tsinghua.edu.cn/simple/
pip install easyocr 
python 复制代码
import easyocr
import os
import cv2
import time
from paddleocr import PaddleOCR

def get_photo_paths(photos_folder_path):
    return [ f"{photos_folder_path}{os.sep}{photo_name}" for photo_name in os.listdir(photos_folder_path) if ".PNG" in photo_name]


def preprocess_image(image):
    # 转为灰度图
    gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
    # 二值化
    _, binary = cv2.threshold(gray, 150, 255, cv2.THRESH_BINARY + cv2.THRESH_OTSU)
    # 去噪声
    denoised = cv2.medianBlur(binary, 3)
    return denoised

def image_cutting(image_path
                  ,up_ratio=0.03
                  ,down_ratio=0.13):
    image = cv2.imread(image_path)
    # 计算切割的宽度和高度
    width = int(image.shape[1] * 1)
    down = int(image.shape[0] * down_ratio)
    up = int(image.shape[0] * up_ratio)
    # 切割图片
    cropped_image = image[up:down, :width]
    # 显示切割后的图片
    # cv2.imshow('Cropped Image', cropped_image)
    preprocessed_image = preprocess_image(cropped_image)
    # plt.imshow(preprocessed_image)
    return preprocessed_image


def get_key_fields_from_easyorc(image,keyword=''):
    res = Ereader.readtext(image)
    texts = []
    if keyword != '':
        for fields in res:
            boundaries = fields[0] ## 边界坐标
            text = fields[1] ## 文字内容
            if keyword in text:
                return {True:text}
            else:
                texts.append(text)
        return {False:texts}
    else:
        text = res[0][1]
        return text 
    

def get_key_fields_from_PaddleOCR(image,keyword=''):
    res = Preader.ocr(image, cls=True)[0]
    texts = []
    if keyword != '':
        for fields in res:
            boundaries = fields[0] ## 边界坐标
            text = fields[1][0] ## 文字内容
            if keyword in text:
                return {True:text}
            else:
                texts.append(text)
        return {False:texts}
    else:
        text = res[0][1][0]
        return {True:text} 
    # res = Preader.ocr(image, cls=True)[0]
    # texts = []
    # for fields in res:
    #     boundaries = fields[0] ## 边界坐标
    #     text = fields[1][0] ## 文字内容
    #     if keyword in text:
    #         return {True:text}
    #     else:
    #         texts.append(text)
    # return {False:texts}

def time_counter(begin_time, end_time):
    # 根据传入的时间计算,通过run_time.round()函数取整
    runtime = round(end_time - begin_time)
    # 计算时分秒
    hour = runtime // 3600
    minute = (runtime - 3600 * hour) // 60
    second = runtime - 3600 * hour - 60 * minute
    # 输出
    return f'用时:{hour}小时{minute}分钟{second}秒'

def change_wechat_photo_name_logic(photos_folder_path
                                   ,keyword = '22级实习-'
                                   ,up_ratio=0.05
                                   ,down_ratio=0.13
                                   ,model_name = "paddleorc"):
    

    if model_name == "paddleorc":
        global Preader
        Preader = PaddleOCR(use_angle_cls=True,det=False, lang="ch")
    elif model_name == "easyorc":
        global Ereader
        Ereader = easyocr.Reader(['ch_sim','en']) # 只需要运行一次就可以将模型加载到内存中
    photo_paths = get_photo_paths(photos_folder_path)
    m = len(photo_paths)
    t = 0
    start_time = time.time()
    for photo_path in photo_paths:
    
        start_time_of_each_step = time.time()
        try:
            preprocessed_image = image_cutting(photo_path,up_ratio,down_ratio)
       
            if model_name == "easyorc":
                wechat_name = get_key_fields_from_easyorc(preprocessed_image,keyword) 
            elif model_name == "paddleorc":
                wechat_name = get_key_fields_from_PaddleOCR(preprocessed_image,keyword)
                
            if keyword != "":     
                split_wechate_name = wechat_name[True].split('-')
                friend_name = split_wechate_name[-1]
            else:
                friend_name = wechat_name[True]
            
            # folder_path = os.path.split(photo_path)[0]
            new_photo_path = os.path.join(photos_folder_path,f"{friend_name}.PNG")
            os.rename(photo_path,new_photo_path)
            t += 1
            end_time_of_each_step = time.time()
            time_for_this_time = time_counter(start_time_of_each_step,end_time_of_each_step)
            print(f"当前好友名为{friend_name},是第{t}个,完成{t/m*100}%,{time_for_this_time}")
        except Exception as e:
            print(f"\033[31m当前文件为:【{photo_path}】\n,错误:{e}\033[0m")
            # print(f"当前好友名为{friend_name},错误:{e}")
    print(f"总计{time_counter(start_time,end_time_of_each_step)},完成{t}个")    

if __name__ == '__main__':
    photos_folder_path = "/Users/magu/Downloads/淮职课程准备/1就业材料/就业实习留存材料/回访截图/张晶实习回访(11月1日-11月30)"
    change_wechat_photo_name_logic(photos_folder_path,keyword = '22级实习-',up_ratio=0.05,down_ratio=0.101)
    #change_wechat_photo_name_logic(photos_folder_path,keyword = '',up_ratio=0.05,down_ratio=0.10)


    
相关推荐
云道轩4 分钟前
使用Docker在Rocky Linux 9.5上在线部署LangFlow
linux·人工智能·docker·容器·langflow
POLOAPI22 分钟前
从模型到生产:AI 大模型落地工程与效率优化实践
人工智能·gpt·gemini
谷歌上搜百度26 分钟前
LLM并非“万能钥匙”——深度解析大语言模型的本质与边界
人工智能·llm
王者鳜錸33 分钟前
PYTHON从入门到实践-15数据可视化
开发语言·python·信息可视化
杨航 AI38 分钟前
ADB+Python控制(有线/无线) Scrcpy+按键映射(推荐)
开发语言·python·adb
Wendy144141 分钟前
【图像掩膜】——图像预处理(OpenCV)
人工智能·opencv·计算机视觉
机器之心1 小时前
开启RL Scaling新纪元,siiRL开源:完全分布式强化学习框架,支持超千卡规模高效训练
人工智能
GISer_Jing1 小时前
Coze:字节跳动AI开发平台功能和架构解析
javascript·人工智能·架构·开源
郝学胜-神的一滴1 小时前
Python defaultdict 的强大之处:告别繁琐的字典键检查: Effective Python 第17条
开发语言·python·程序人生
我有一计3331 小时前
【算法笔记】5.LeetCode-Hot100-矩阵专项
人工智能·算法·程序员