Improve deduplication of lines; intermediate commit for debugging.

src/stratigraphy/line_detection.py

@@ -54,9 +54,7 @@ def detect_lines_lsd(page: fitz.Page, scale_factor=2, lsd_params=None) -> ArrayL
 
     # Detect lines in the image
     lines = lsd.detect(gray)[0]
-    converted_lines = [line_from_array(line, scale_factor) for line in lines]
-    deduplicated_lines = deduplicate_lines(converted_lines)
-    return deduplicated_lines
+    return [line_from_array(line, scale_factor) for line in lines]
 
 
 def extract_lines(page: fitz.Page, line_detection_params: dict) -> list[Line]:
@@ -75,6 +73,7 @@ def extract_lines(page: fitz.Page, line_detection_params: dict) -> list[Line]:
         scale_factor=line_detection_params["pdf_scale_factor"],
     )
     lines = drop_vertical_lines(lines, threshold=line_detection_params["vertical_lines_threshold"])
+    lines = deduplicate_lines(lines)
     merging_params = line_detection_params["line_merging_params"]
     if merging_params["use_clustering"]:
         lines = merge_parallel_lines_approximately(

src/stratigraphy/util/dataclasses.py

@@ -71,6 +71,16 @@ def remove(self, line_index: str):
             del self.hashmap[line_index]
 
     def add(self, line: Line) -> str:
-        key = uuid.uuid4().hex
-        self.hashmap[key] = line
-        return key
+        if not self._check_if_present(line):
+            key = uuid.uuid4().hex
+            self.hashmap[key] = line
+            return key
+        else:
+            logger.warning("Line already present in IndexedLines.")
+            return None
+
+    def _check_if_present(self, line: Line) -> bool:
+        return any(
+            value.start.distance_to(line.start) < 0.1 and value.end.distance_to(line.end) < 0.1
+            for _key, value in self.hashmap.items()
+        )

src/stratigraphy/util/geometric_line_utilities.py

@@ -37,8 +37,10 @@ def deduplicate_lines(lines: list[Line]) -> list[Line]:
     return deduplicated_lines
 
 
-def _check_if_present(lines, line: Line) -> bool:
-    return any(value.start == line.start and value.end == line.end for value in lines)
+def _check_if_present(lines: list[Line], line: Line) -> bool:
+    return any(
+        value.start.distance_to(line.start) < 0.1 and value.end.distance_to(line.end) < 0.1 for value in lines
+    )  # we are on a pixel grid and 0.1 is a reasonable threshold
 
 
 def drop_vertical_lines(lines: list[Line], threshold: float = 0.1) -> ArrayLike:
@@ -476,6 +478,7 @@ def merge_parallel_lines_quadtree(lines: list[Line], tol: int, angle_threshold:
                             merged_any = True
                         continue
     if merged_any:
+        print("Starting recursion.")
         return merge_parallel_lines_quadtree(
             list(indexed_lines.hashmap.values()), tol=tol, angle_threshold=angle_threshold
         )