From 74b5d33c86c9c1891ab6222cb5e2fa55ae974e3d Mon Sep 17 00:00:00 2001
From: Claude Agent <claude-agent@romfast.ro>
Date: Thu, 16 Apr 2026 07:08:28 +0000
Subject: [PATCH] fix(vision): connected-components for rightmost-dot detection
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Dots on M2D MAPS strip are so close that anti-aliased edge pixels bridge
adjacent dots column-counts → the previous walk-left approach merged the
entire strip into one cluster and picked its midpoint.

Connected components (8-connectivity) treats each dot as a separate blob
even when antialiased edges touch. We pick the blob with the largest
right-edge, then return its centroid. Robust, O(pixels), one opencv call.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
---
 src/atm/vision.py | 41 +++++++++++++++++++----------------------
 1 file changed, 19 insertions(+), 22 deletions(-)

diff --git a/src/atm/vision.py b/src/atm/vision.py
index fa9f378..2a1016c 100644
--- a/src/atm/vision.py
+++ b/src/atm/vision.py
@@ -101,30 +101,27 @@ def find_rightmost_dot(
     """
     bgr_bg = np.array([bg_rgb[2], bg_rgb[1], bg_rgb[0]], dtype=np.float32)
     diff = np.linalg.norm(roi_img.astype(np.float32) - bgr_bg, axis=2)
-    mask = diff > bg_tol  # True = non-background
-    _h, w = mask.shape
-    col_counts = mask.sum(axis=0)
+    mask = (diff > bg_tol).astype(np.uint8)
 
-    right_col = None
-    for x in range(w - 1, -1, -1):
-        if col_counts[x] >= min_cluster_px:
-            right_col = x
-            break
-    if right_col is None:
+    # Connected components → one component per dot. Anti-aliasing bridges
+    # between adjacent dots are small enough that 8-connectivity still
+    # separates them cleanly.
+    n_labels, _labels, stats, centroids = cv2.connectedComponentsWithStats(
+        mask, connectivity=8,
+    )
+    best_idx: int | None = None
+    best_right = -1
+    for i in range(1, n_labels):  # skip background (label 0)
+        if int(stats[i, cv2.CC_STAT_AREA]) < min_cluster_px:
+            continue
+        right = int(stats[i, cv2.CC_STAT_LEFT] + stats[i, cv2.CC_STAT_WIDTH])
+        if right > best_right:
+            best_right = right
+            best_idx = i
+    if best_idx is None:
         return None
-
-    # Walk LEFT until a gap to find the cluster's left edge.
-    left_col = right_col
-    for x in range(right_col - 1, -1, -1):
-        if col_counts[x] < min_cluster_px:
-            break
-        left_col = x
-
-    cx = (left_col + right_col) // 2
-    cluster = mask[:, left_col:right_col + 1]
-    ys = np.where(cluster.any(axis=1))[0]
-    cy = int(ys.mean()) if ys.size > 0 else 0
-    return (cx, cy)
+    cx, cy = centroids[best_idx]
+    return (int(cx), int(cy))
 
 
 def pixel_rgb(roi_img: np.ndarray, x: int, y: int, box: int = 3) -> tuple[int, int, int]: