fix:ocr algorithm issue with repetitive subtitle

detect_with_ocr.py

@@ -406,6 +406,7 @@ def process_video(video_path: str, begin: float, end: float, book_path: str, she
     # res是在视频遍历过程中获取的字幕文件，不掺杂对旁白的分析
     res = []
     cnt = 0
+    cur_time1 = 0
     start_time = 0
     end_time = 0
     video.set(cv2.CAP_PROP_POS_MSEC, begin * 1000)
@@ -436,6 +437,7 @@ def process_video(video_path: str, begin: float, end: float, book_path: str, she
             break
         # 每秒取4帧画面左右
         # TODO 取帧算法优化
+        fl = False
         if cnt % int(fps / 4) == 0:
             # 更新当前工程的检测进度
             if pre_state is None:
@@ -453,7 +455,30 @@ def process_video(video_path: str, begin: float, end: float, book_path: str, she
                 subTitle = normalize(subTitle)
                 if len(subTitle) == 0:
                     subTitle = None
-
+            if fl:
+                if subTitle is None:
+                    continue
+                else :
+                    if string_similar(lastSubTitle, subTitle) < 0.7:
+                        end_time = cur_time1
+                        res.append([start_time, end_time, lastSubTitle])
+                        if (len(res) == 1 and res[-1][0] - last_time >= 1) or (len(res) > 1 and res[-1][0] - res[-2][1]) >= 1:
+                            print('--------------------------------------------------')
+                            recommend_lens = int((res[-1][0] - last_time) * normal_speed) if len(res) == 1 else int(
+                                (res[-1][0] - res[-2][1]) * normal_speed)
+                            # write_to_sheet(book_path, sheet_name, ['', '', '', '插入旁白，推荐字数为%d' % recommend_lens])
+                            # add_to_list(mainWindow, "旁白", ['', '', '', '插入旁白，推荐字数为%d' % recommend_lens])
+                            add_to_list(mainWindow, "旁白", ['', '', '', '%d' % recommend_lens],ocr_h)
+                        print(start_time, end_time, lastSubTitle)
+
+                        # write_to_sheet(book_path, sheet_name, [round(start_time, 2), round(end_time, 2), lastSubTitle, ''])
+                        add_to_list(mainWindow, "字幕", [round(start_time, 3), round(end_time, 3), lastSubTitle, ''],ocr_h)
+                        print(">>>>>>subtitle,ocr_h2:" + str(lastSubTitle) + ">>>" + str(ocr_h))
+                        start_time = cur_time
+                    else :
+                        lastSubTitle = subTitle if conf > lastConf else lastSubTitle
+                        lastConf = max(lastConf, conf)
+                    fl = False
             # 第一次找到字幕
             if lastSubTitle is None and subTitle is not None:
                 if cur_ocr_h != None:
@@ -462,6 +487,9 @@ def process_video(video_path: str, begin: float, end: float, book_path: str, she
 
             # 字幕消失
             elif lastSubTitle is not None and subTitle is None:
+                fl = False
+                cur_time1 = cur_time
+                continue
                 end_time = cur_time
                 res.append([start_time, end_time, lastSubTitle])
                 if (len(res) == 1 and res[-1][0] - last_time >= 1) or (len(res) > 1 and res[-1][0] - res[-2][1]) >= 1: