使用Python开发一个图像标注与OCR识别工具

self.image_files = [f for f in os.listdir(folder_path) 
                    if f.lower().endswith(('.png', '.jpg', '.jpeg', '.bmp', '.tiff'))]

file_path = os.path.join(self.folder_path, filename)
self.original_image = cv2.imdecode(np.fromfile(file_path, dtype=np.uint8), cv2.IMREAD_COLOR)
self.original_image = cv2.cvtColor(self.original_image, cv2.COLOR_BGR2RGB)

def on_mouse_down(self, event):
    """鼠标按下开始绘制矩形"""
    if self.displayed_image is not None:
        self.start_point = event.GetPosition()
        self.drawing = True

def on_mouse_move(self, event):
    """鼠标移动时更新矩形"""
    if self.drawing and self.displayed_image is not None:
        self.image_panel.Refresh()

​​​​​​​def on_mouse_up(self, event):
    """鼠标松开完成矩形绘制"""
    if self.drawing and self.displayed_image is not None:
        end_point = event.GetPosition()
        # 确保矩形方向正确
        x1 = min(self.start_point.x, end_point.x)
        y1 = min(self.start_point.y, end_point.y)
        x2 = max(self.start_point.x, end_point.x)
        y2 = max(self.start_point.y, end_point.y)
        orig_rect = self.convert_to_original_coords((x1, y1, x2, y2))
        self.rectangles.append(orig_rect)
        self.drawing = False
        self.image_panel.Refresh()

text = pytesseract.image_to_string(
    pil_image, 
    lang='chi_sim',  # 中文简体
    config='--psm 6 --oem 3'  # 更精确的文本块处理
)

if ocr_results:
    df = pd.DataFrame({'识别区域': range(1, len(ocr_results) + 1), '识别文本': ocr_results})
    output_path = os.path.join(self.folder_path, f'{self.current_filename}_ocr_results.xlsx')
    df.to_excel(output_path, index=False, engine='openpyxl')

annotations_data = {
    'filename': self.current_filename,
    'rectangles': self.rectangles
}
json_path = os.path.join(self.folder_path, f'{self.current_filename}_annotations.json')
with open(json_path, 'w', encoding='utf-8') as f:
    json.dump(annotations_data, f)

with open(json_path, 'r', encoding='utf-8') as f:
    annotations_data = json.load(f)
if annotations_data['filename'] == self.current_filename:
    self.rectangles = annotations_data['rectangles']

cropped = self.current_image[y1:y2, x1:x2].copy()

file_list_sizer = wx.BoxSizer(wx.VERTICAL)
self.file_listbox = wx.ListBox(panel, choices=self.image_files, style=wx.LB_SINGLE)
file_list_sizer.Add(self.file_listbox, 1, wx.EXPAND | wx.ALL, 10)