Merge remote-tracking branch 'origin/bspline_integration'

Author: cindygr

EvalRoutines/ReadMe.txt eval_routines/ReadMe.txt

@@ -0,0 +1,38 @@
+#!/usr/bin/env python3
+import json
+import numpy as np
+
+class fit_params:
+    def __init__(
+        self,
+        json_string = None,
+    ):
+        self.load_json(json_string)
+
+    def is_param(self, str):
+        return str in self.__dict__.keys()
+
+    def get_param(self, str):
+        return self.__dict__[str]
+    
+    def add_param(self, str, val):
+        self.__dict__[str] = val
+
+    def update_param(self, str, val):
+        self.__dict__[str] = val
+
+    def json_dump(self):
+        return json.dumps(self.__dict__, indent=4)
+
+    def load_json(self, json_str: str):
+        self.__dict__ = json.loads(json_str)
+        for key in self.__dict__.keys():
+            if isinstance(self.__dict__[key], list):
+                self.__dict__[key] = np.array(self.__dict__[key])
+
+if __name__ == "__main__":
+    fp = fit_params(
+        1.0, residuals=np.array([0.6, 0.3, 0.2]), ts=np.array([0.2, 0.7, 1.0])
+    )
+    print(fp.json_dump())
+    pass

FitRoutines/ReadMe.txt fit_routines/ReadMe.txt

@@ -0,0 +1,43 @@
+#!/usr/bin/env python3
+from numpy.polynomial import polynomial as P
+from numpy.polynomial import Polynomial
+import numpy as np
+import matplotlib.pyplot as plt
+
+lut = [      [1],         
+            [1,1],        
+           [1,2,1],       
+          [1,3,3,1],      
+         [1,4,6,4,1],     
+        [1,5,10,10,5,1],  
+       [1,6,15,20,15,6,1]]
+
+def binomial(n : int, k: int):
+  while n >= len(lut):
+    s = len(lut)
+    nextRow = [1] * s
+    for i in range(1, s):
+      nextRow[i] = lut[s-1][i-1] + lut[s-1][i]
+    nextRow.append(1)
+    lut.append(nextRow)
+  return lut[n][k]
+
+def bezier_basis(n):
+  basis = []
+  fig, ax = plt.subplots()
+  for k in range (0, n+1):
+    print(f"n: {n} k: {k}")
+    term1 = P.polypow([1, -1], n-k)
+    term2 = P.polypow([0, 1], k)
+    mult = P.polymul([binomial(n,k)], term1)
+    mult2 = P.polymul(mult, term2)
+    p = Polynomial(coef=mult2)
+    m = p.linspace(100, [0, 1])
+    ax.plot(m[0], m[1])
+    basis.append(mult2)
+  print(basis)
+  plt.show()
+  return basis
+
+if __name__ == "__main__":
+    bezier_basis(12)

fit_routines/__init__.py

@@ -0,0 +1,50 @@
+#!/usr/bin/env python3
+import os
+import cv2
+import sys
+import numpy as np
+
+""" Create mask images assuming data formatted as "labelled_id.png" is available at 
+one folder level below the path passed to the script
+"""
+
+LABELS = {
+    (69, 244, 139): "spur",
+    (255, 255, 255): "trunk",
+    (87, 192, 255): "branch",
+    (0, 0, 244): "leader",
+}
+
+# (0 0 0): background tree,
+# (255, 90, 67): sky,
+# (255 0 0) poles,
+# (0 0 244) leader,
+
+def generate_masks(files, output_dir):
+    for file in files:
+        img = cv2.imread(file)
+        # black mask
+        curr_mask = np.zeros(img.shape[:2], dtype=np.uint8)
+        for color in LABELS.keys():
+            color_arr = np.array(color)
+            
+            mask = cv2.inRange(img, color_arr - 10, color_arr + 10)
+            # or with previous mask
+            curr_mask = cv2.bitwise_or(mask, curr_mask)
+            start_index = file.find("labelled") + len("labelled")
+            # mask_name = f"mask_{LABELS[color]}_{file[start_index:]}"
+            # print(f"Saving mask {mask_name} in {output_dir}")
+            # cv2.imwrite(os.path.join(output_dir, mask_name), mask)
+        mask_name = f"mask_all_{file[start_index:]}"
+        cv2.imwrite(os.path.join(output_dir, mask_name), curr_mask)
+        print(f"Saving mask {mask_name} in {output_dir}")
+
+if __name__ == '__main__':
+    dir_path = sys.argv[1]
+    output_dir = dir_path
+    os.makedirs(output_dir, exist_ok=True)
+    for subfolder in os.listdir(dir_path):
+        subdir_path = os.path.join(dir_path, subfolder)
+        files = [os.path.join(subdir_path, x) for x in os.listdir(subdir_path) if x.endswith(".png") and x.startswith("labelled")]
+        generate_masks(files, os.path.join(output_dir, subfolder))
+

fit_routines/b_spline_curve_fit.py

@@ -0,0 +1,16 @@
+def get_contiguous_distance(pts, matches, vec):
+    current_start = None
+    dist = 0
+    for i, match in enumerate(matches):
+        if current_start is None and match:
+            current_start = i
+        elif current_start is not None and not match:
+            offsets = pts[current_start + 1 : i] - pts[current_start : i - 1]
+            dist += (offsets * vec).sum()
+            current_start = None
+    if current_start is not None:
+        i = len(matches)
+        offsets = pts[current_start + 1 : i] - pts[current_start : i - 1]
+        dist += (offsets * vec).sum()
+
+    return dist