|
@@ -12,7 +12,7 @@ from fuzzywuzzy import fuzz
|
|
|
class OCRResultComparator:
|
|
class OCRResultComparator:
|
|
|
def __init__(self):
|
|
def __init__(self):
|
|
|
self.differences = []
|
|
self.differences = []
|
|
|
- self.similarity_threshold = 85
|
|
|
|
|
|
|
+ self.similarity_threshold = 95
|
|
|
self.max_paragraph_window = 6
|
|
self.max_paragraph_window = 6
|
|
|
self.table_comparison_mode = 'standard' # 新增：表格比较模式
|
|
self.table_comparison_mode = 'standard' # 新增：表格比较模式
|
|
|
self.header_similarity_threshold = 80 # 表头相似度阈值
|
|
self.header_similarity_threshold = 80 # 表头相似度阈值
|
