影刀RPA分享第91期|手机自动化判断元素抓取新思路 通讯录采集图像比对法

教程如下~~~

https://www.bilibili.com/video/BV1LtaJzmEdz/

指令作用

给你一张图像 在给你一个文件来，文件夫有很多图像你给我判断这个图片是不是已经在文件表中，不是用md5判断 用像素相似度

代码如下：

# 使用此指令前，请确保安装必要的Python库，例如使用以下命令安装：
# pip install opencv-python numpy pillow

import os
import cv2
import numpy as np
from PIL import Image

from typing import *
try:
    from xbot.app.logging import trace as print
except:
    from xbot import print


def check_image_similarity(target_image, images_folder, similarity_threshold=0.95):
    """
    title: 图像相似度检测
    description: 检查目标图像 % target_image % 是否与文件夹 % images_folder % 中的任何图像相似，基于像素相似度而非MD5。
    inputs: 
        - target_image (file): 需要检查的目标图像文件路径，eg: "target.jpg"
        - images_folder (folder): 包含多个图像的文件夹路径，eg: "images_folder/"
        - similarity_threshold (float): 相似度阈值，范围0-1，默认0.95，eg: "0.95"
    outputs: 
        - result (dict): 包含是否找到相似图像及相似图像信息的字典，eg: "{'found': True, 'similar_image': 'image1.jpg', 'similarity': 0.98}"
    """
    
    # 检查输入参数
    if not os.path.isfile(target_image):
        raise FileNotFoundError(f"目标图像文件不存在: {target_image}")
    
    if not os.path.isdir(images_folder):
        raise FileNotFoundError(f"图像文件夹不存在: {images_folder}")
    
    if not 0 <= similarity_threshold <= 1:
        raise ValueError("相似度阈值必须在0到1之间")
    
    # 读取目标图像
    def _read_image(image_path):
        try:
            # 使用PIL读取图像，然后转换为OpenCV格式，这样可以更好地处理中文路径
            img_pil = Image.open(image_path)
            img_np = np.array(img_pil)
            # 如果图像是RGB格式，转换为BGR (OpenCV使用BGR)
            if len(img_np.shape) == 3 and img_np.shape[2] == 3:
                img_np = cv2.cvtColor(img_np, cv2.COLOR_RGB2BGR)
            return img_np
        except Exception as e:
            raise ValueError(f"读取图像时出错: {str(e)}")
    
    # 计算两张图像的相似度
    def _calculate_similarity(img1, img2):
        try:
            # 调整图像大小以便比较
            height = 200  # 可以根据需要调整
            width = 200
            img1_resized = cv2.resize(img1, (width, height))
            img2_resized = cv2.resize(img2, (width, height))
            
            # 确保图像是3通道的，如果不是则转换
            if len(img1_resized.shape) == 2:
                img1_resized = cv2.cvtColor(img1_resized, cv2.COLOR_GRAY2BGR)
            if len(img2_resized.shape) == 2:
                img2_resized = cv2.cvtColor(img2_resized, cv2.COLOR_GRAY2BGR)
            
            # 转换为灰度图像
            img1_gray = cv2.cvtColor(img1_resized, cv2.COLOR_BGR2GRAY)
            img2_gray = cv2.cvtColor(img2_resized, cv2.COLOR_BGR2GRAY)
            
            # 使用均方误差(MSE)作为相似度度量
            err = np.sum((img1_gray.astype("float") - img2_gray.astype("float")) ** 2)
            err /= float(img1_gray.shape[0] * img1_gray.shape[1])
            
            # 将MSE转换为相似度分数 (0-1)
            similarity = 1 - (err / 255**2)
            return similarity
        except Exception as e:
            raise ValueError(f"计算相似度时出错: {str(e)}")
    
    # 主处理逻辑
    try:
        target_img = _read_image(target_image)
    except Exception as e:
        raise ValueError(f"无法读取目标图像: {target_image}, 错误: {str(e)}")
    
    result = {
        'found': False,
        'similar_image': None,
        'similarity': 0
    }
    
    # 获取文件夹中的所有图像文件
    image_extensions = ['.jpg', '.jpeg', '.png', '.bmp', '.gif', '.tiff']
    try:
        image_files = [f for f in os.listdir(images_folder) 
                      if os.path.isfile(os.path.join(images_folder, f)) and 
                      os.path.splitext(f.lower())[1] in image_extensions]
    except Exception as e:
        raise ValueError(f"无法读取文件夹内容: {images_folder}, 错误: {str(e)}")
    
    if not image_files:
        print(f"警告: 文件夹 {images_folder} 中没有找到图像文件")
        return result
    
    # 比较目标图像与文件夹中的每张图像
    highest_similarity = 0
    most_similar_image = None
    
    for image_file in image_files:
        image_path = os.path.join(images_folder, image_file)
        try:
            img = _read_image(image_path)
            similarity = _calculate_similarity(target_img, img)
            
            if similarity > highest_similarity:
                highest_similarity = similarity
                most_similar_image = image_file
                
            if similarity >= similarity_threshold:
                result['found'] = True
                result['similar_image'] = image_file
                result['similarity'] = similarity
                return result
                
        except Exception as e:
            print(f"处理图像 {image_file} 时出错: {str(e)}")
            continue
    
    # 如果没有找到超过阈值的相似图像，返回最相似的一个
    if most_similar_image:
        result['similar_image'] = most_similar_image
        result['similarity'] = highest_similarity
    
    return result

影刀指令说明