# Pre registration, start-up!

cellbin2/contrib/alignment/template_00pt.py

@@ -44,7 +44,7 @@ def registration_image(
 
         result = image.trans_image(
             scale=[1 / self._scale_x, 1 / self._scale_y],
-            rotate=-self._rotation,
+            rotate=self._rotation,
             rot90=self.rot90,
             offset=self.offset,
             dst_size=self._register_shape,
@@ -87,6 +87,7 @@ def align_stitched(self, moving_image: ChipFeature, ):
         )
 
         transformed_feature.set_point00(moving_image.point00)
+        transformed_feature.set_anchor_point(moving_image.anchor_point)
         transformed_feature.chip_box.set_chip_box(chip_points)
         transformed_feature.set_template(
             np.concatenate(
@@ -126,7 +127,10 @@ def align_transformed(self, moving_image: ChipFeature):
         _points = self.get_lt_zero_point(_points)
         _points = _points[(_points[:, 0] > 0) & (_points[:, 1] > 0)]
 
-        px, py = sorted(_points.tolist(), key=lambda x: x[0] + x[1])[0] + chip_box_points[0]
+        px, py = sorted(
+            _points.tolist(),
+            key=lambda x: np.abs(x[0] - moving_image.anchor_point[0]) + np.abs(x[1] - moving_image.anchor_point[1])
+        )[0] + chip_box_points[0]
 
         self._offset = [moving_image.point00[0] - px, moving_image.point00[1] - py]
 

cellbin2/contrib/param.py

@@ -16,6 +16,7 @@ def __init__(self):
         self._point00: Tuple[int, int] = (0, 0)  # xy,相对于芯片而不是矩阵的位置坐标
         self._mat: Union[str, np.ndarray, CBImage] = ''
         # self.ref: List[List, List] = [[]]
+        self._anchor_point: Tuple[int, int] = (0, 0)  # xy, 用于配准前置的辅助锚点
 
     @property
     def chip_box(self, ):
@@ -28,6 +29,14 @@ def set_point00(self, points):
         if isinstance(points, tuple) and len(points) == 2:
             self._point00 = points
 
+    def set_anchor_point(self, points):
+        if isinstance(points, tuple) and len(points) == 2:
+            self._anchor_point = points
+
+    @property
+    def anchor_point(self, ):
+        return self._anchor_point
+
     @property
     def point00(self, ):
         return self._point00

cellbin2/modules/extract/image_extract.py

@@ -257,6 +257,7 @@ def _pre_registration(self, ):
         moving_image.set_template(self._channel_image.stitched_template_info)
         moving_image.set_chip_box(self._channel_image.box_info)
         moving_image.set_point00(self._param_chip.zero_zero_point)
+        moving_image.set_anchor_point(tuple(self._param_chip.zero_zero_chip_point))
 
         res = template_00pt.template_00pt_align(moving_image=moving_image,
                                                 ref=self._param_chip.fov_template,
@@ -266,6 +267,8 @@ def _pre_registration(self, ):
         with open(os.path.join(self.output_path, 'register_00pt.txt'), 'w') as f:
             f.writelines(f"offset: {res.offset} \n")
 
+        # res.register_mat.write(os.path.join(self.output_path, "register_00pt.tif"))
+
         # self._channel_image.update_registration(res)
 
     @process_decorator('GiB')

cellbin2/utils/stereo_chip.py

@@ -148,8 +148,8 @@ def set_zero_point_info(self, ):
                            points_finish.transpose(1, 0).tolist())
         _x = np.array(x_set)[1:] - np.array(x_set)[:-1]
         _y = np.array(y_set)[1:] - np.array(y_set)[:-1]
-        zero_x = x_set[np.where(_x == self.fov_template[0][0])[0][0] + 2]
-        zero_y = y_set[np.where(_y == self.fov_template[0][0])[0][0] + 2]
+        zero_x = x_set[np.where(_x == max(self.fov_template[0]))[0][0] + 1]
+        zero_y = x_set[np.where(_x == max(self.fov_template[0]))[0][0] + 1]
 
         # 补充00点距芯片角距离
         index = np.where(((points_finish[:, 0] == zero_x) & (points_finish[:, 1] == zero_y)) == True)[0][0]