Files
video_detect/service/ocr_service.py

237 lines
8.5 KiB
Python
Raw Normal View History

import time
2025-09-30 17:17:20 +08:00
import numpy as np
# 修复np.int已弃用的问题
if not hasattr(np, 'int'):
np.int = int
from paddleocr import PaddleOCR
from service.sensitive_service import get_all_sensitive_words
# AC自动机节点定义
class AhoNode:
def __init__(self):
self.children = {} # 子节点映射(字符->节点)
self.fail = None # 失败指针类似KMP的next数组
self.is_end = False # 标记是否为某个模式串的结尾
self.word = None # 存储当前结尾对应的完整违禁词
# AC自动机实现多模式字符串匹配
class AhoCorasick:
def __init__(self):
self.root = AhoNode() # 根节点
def add_word(self, word):
"""添加违禁词到Trie树"""
if not isinstance(word, str) or not word.strip():
return # 过滤无效词
node = self.root
for char in word:
if char not in node.children:
node.children[char] = AhoNode()
node = node.children[char]
node.is_end = True
node.word = word # 记录完整词
def build_fail(self):
"""构建失败指针BFS遍历"""
queue = []
# 根节点的子节点失败指针指向根节点
for child in self.root.children.values():
child.fail = self.root
queue.append(child)
# BFS处理其他节点
while queue:
current_node = queue.pop(0)
# 遍历当前节点的所有子节点
for char, child in current_node.children.items():
# 寻找失败指针目标节点
fail_node = current_node.fail
while fail_node is not None and char not in fail_node.children:
fail_node = fail_node.fail
# 确定失败指针指向
child.fail = fail_node.children[char] if (fail_node and char in fail_node.children) else self.root
queue.append(child)
def match(self, text):
"""匹配文本中所有出现的违禁词(去重)"""
result = set()
node = self.root
for char in text:
# 沿失败链查找可用节点
while node is not None and char not in node.children:
node = node.fail
# 重置到根节点(如果没找到)
node = node.children[char] if (node and char in node.children) else self.root
# 收集所有匹配的违禁词(包括失败链上的)
temp = node
while temp != self.root:
if temp.is_end:
result.add(temp.word)
temp = temp.fail
return list(result)
# 全局变量
2025-09-30 17:17:20 +08:00
_ocr_engine = None
_ac_automaton = None # 替换原有的_forbidden_words集合
2025-09-30 17:17:20 +08:00
_conf_threshold = 0.5
2025-09-30 17:17:20 +08:00
def set_forbidden_words(new_words):
"""更新违禁词使用AC自动机存储"""
global _ac_automaton
2025-09-30 17:17:20 +08:00
if not isinstance(new_words, (set, list, tuple)):
raise TypeError("新违禁词必须是集合、列表或元组类型")
# 初始化AC自动机并添加有效词
_ac_automaton = AhoCorasick()
valid_words = [word for word in new_words if isinstance(word, str) and word.strip()]
for word in valid_words:
_ac_automaton.add_word(word.strip())
# 构建失败指针(关键步骤)
_ac_automaton.build_fail()
print(f"已通过函数更新违禁词,当前数量: {len(valid_words)}")
2025-09-30 17:17:20 +08:00
def load_forbidden_words():
"""从敏感词服务加载违禁词并初始化AC自动机"""
global _ac_automaton
2025-09-30 17:17:20 +08:00
try:
sensitive_words = get_all_sensitive_words() # 保持原接口不变返回list[str]
_ac_automaton = AhoCorasick()
# 添加所有有效敏感词
valid_words = [word for word in sensitive_words if isinstance(word, str) and word.strip()]
for word in valid_words:
_ac_automaton.add_word(word.strip())
# 构建失败指针
_ac_automaton.build_fail()
print(f"加载的违禁词数量: {len(valid_words)}")
return True
2025-09-30 17:17:20 +08:00
except Exception as e:
print(f"Forbidden words load error: {e}")
return False
def init_ocr_engine():
"""初始化OCR引擎和违禁词自动机"""
2025-09-30 17:17:20 +08:00
global _ocr_engine
try:
_ocr_engine = PaddleOCR(
use_angle_cls=True,
lang="ch",
show_log=False,
use_gpu=True,
max_text_length=1024
)
load_result = load_forbidden_words()
if not load_result:
print("警告:违禁词加载失败,可能影响检测功能")
print("OCR引擎初始化完成")
return True
except Exception as e:
print(f"OCR引擎初始化错误: {e}")
_ocr_engine = None
return False
def detect(frame, conf_threshold=0.8):
"""检测帧中的文本是否包含违禁词拆分OCR和匹配时间"""
2025-09-30 17:17:20 +08:00
print("开始进行OCR检测...")
total_start = time.time() # 总耗时开始
ocr_time = 0.0 # OCR及结果解析耗时
match_time = 0.0 # 违禁词匹配耗时
2025-09-30 17:17:20 +08:00
try:
if not _ocr_engine or not _ac_automaton:
return (False, "OCR引擎或违禁词库未初始化")
# 1. OCR识别及结果解析阶段
ocr_start = time.time()
2025-09-30 17:17:20 +08:00
ocr_res = _ocr_engine.ocr(frame, cls=True)
if not ocr_res or not isinstance(ocr_res, list):
return (False, "无OCR结果")
texts = []
confs = []
# 解析OCR结果
2025-09-30 17:17:20 +08:00
for line in ocr_res:
if line is None:
continue
items_to_process = line if isinstance(line, list) else [line]
2025-09-30 17:17:20 +08:00
for item in items_to_process:
# 过滤坐标类数据
2025-09-30 17:17:20 +08:00
if isinstance(item, list) and len(item) == 4:
is_coordinate = all(isinstance(p, list) and len(p) == 2 and
all(isinstance(c, (int, float)) for c in p)
for p in item)
2025-09-30 17:17:20 +08:00
if is_coordinate:
continue
# 提取文本和置信度
2025-09-30 17:17:20 +08:00
if isinstance(item, tuple) and len(item) == 2:
text, conf = item
if isinstance(text, str) and isinstance(conf, (int, float)):
texts.append(text.strip())
confs.append(float(conf))
continue
if isinstance(item, list) and len(item) >= 2:
text_data = item[1]
if isinstance(text_data, tuple) and len(text_data) == 2:
text, conf = text_data
if isinstance(text, str) and isinstance(conf, (int, float)):
texts.append(text.strip())
confs.append(float(conf))
continue
elif isinstance(text_data, str):
texts.append(text_data.strip())
confs.append(1.0)
continue
ocr_end = time.time()
ocr_time = ocr_end - ocr_start # 计算OCR阶段耗时
2025-09-30 17:17:20 +08:00
if len(texts) != len(confs):
return (False, "OCR结果格式异常")
# 2. 违禁词匹配阶段
match_start = time.time()
2025-09-30 17:17:20 +08:00
vio_words = []
for txt, conf in zip(texts, confs):
if conf < _conf_threshold:
2025-09-30 17:17:20 +08:00
continue
# 用AC自动机匹配当前文本中的所有违禁词
matched_words = _ac_automaton.match(txt)
# 全局去重并保持顺序
for word in matched_words:
2025-09-30 17:17:20 +08:00
if word not in vio_words:
vio_words.append(word)
match_end = time.time()
match_time = match_end - match_start # 计算匹配阶段耗时
2025-09-30 17:17:20 +08:00
has_text = len(texts) > 0
has_violation = len(vio_words) > 0
if not has_text:
return (False, "未识别到文本")
elif has_violation:
return (True, ", ".join(vio_words))
else:
return (False, "未检测到违禁词")
except Exception as e:
print(f"OCR detect error: {e}")
return (False, f"检测错误: {str(e)}")
finally:
# 打印各阶段耗时
total_time = time.time() - total_start
print(f"当前帧耗时明细:")
print(f" OCR识别及解析{ocr_time:.8f}")
print(f" 违禁词匹配:{match_time:.8f}")
print(f" 总耗时:{total_time:.8f}")