added csv extraction

src/experiments/algorithms/data/bgr_stats.csv

@@ -0,0 +1,9 @@
+label,avg_b,avg_g,avg_r,std_b,std_g,std_r,
+accasia,69.59916027137541,93.23599740738842,92.76413721475456,29.650789354560942,28.164331202979742,28.33379470933227,
+berk,87.54178037532186,107.36707160860637,105.9353394079013,41.72180441785215,40.129023307835965,38.64035787139545,
+eik,103.74168566057762,119.53735094782796,117.95327527937933,34.77238255525745,33.829618649660944,32.82833612530793,
+els,138.35231618986808,177.39126510134065,181.7525877978574,51.65085237511483,43.383068821542174,42.175152137736305,
+esdoorn,47.365041621150056,77.05362430457048,69.52487212460042,25.207693719301673,24.628219686465727,22.25080719620302,
+es,63.945103048399794,95.91903641356423,90.20861703612438,20.389426811481062,18.395161770474054,17.716844447273072,
+linde,111.6252480236714,123.48265354152022,129.6323361059727,35.25451067293944,34.66671249273108,33.99378641928277,
+plataan,47.74452753592992,82.96830215527001,85.52863037536183,14.8187819404046,18.13170218888685,19.52480824126827,

src/experiments/algorithms/data/hsv_stats.csv

@@ -0,0 +1,9 @@
+label,avg_b,avg_g,avg_r,std_b,std_g,std_r,
+accasia,35.90440242881845,79.12766192516973,96.03904497370279,10.316069136947837,32.12312271750131,28.112318716352682,
+berk,37.269905856891874,66.77426733032776,109.28940151500194,12.477741919995777,32.92938477640848,39.80902152951404,
+eik,39.76121762093227,41.20602771946309,121.12686669901616,23.03425646554001,25.365001089124593,33.494192142922856,
+els,27.25760648235924,68.91241506727702,182.27155476463224,4.423353762285157,34.149610367865485,42.029268409707505,
+esdoorn,40.44041857512247,131.63818552294646,77.84993286553097,9.535383780285505,39.540648429565636,24.46222201958771,
+es,38.37579404108352,98.88868379328436,96.68345708286985,6.84651057363783,30.56118627938561,18.229878712396474,
+linde,36.8388961321136,41.02522196067761,130.70675471758108,31.06746560545738,22.601057153094764,34.24357220627512,
+plataan,30.208627954766115,123.22600953975079,88.02246907749901,4.833302068080008,20.980674302352114,19.054660495972172,

src/experiments/algorithms/distribution_test.py

@@ -3,25 +3,14 @@ import matplotlib.pyplot as plt
 import cv2
 import os
 import math
-    
-def calcWeightedStd(values, weights):
-    N = len(values)
-    x_bar = np.average(values, weights=weights, axis=0)
-    numerator = 0
-    for i in range(N):
-        numerator += weights[i] * ((values[i] -  x_bar) ** 2)
-    denominator = np.sum(weights)
-    return math.sqrt(numerator / denominator)
 
-def calcNormalFunc(mean, sd, len):
+def calcNormalFunc(mean : float, sd : float, len : int):
     f = np.zeros(len, dtype=np.longdouble)
     
-    # calculate PDF
     for x in range(len):
         exp = (-(x - mean) ** 2)/(2 * sd ** 2)
         f[x] = 1 / math.sqrt(2 * np.pi * sd** 20 ) * (math.exp(exp))
 
-    # normalize PDF
     max = np.amax(f)
     min = np.amin(f)
     for x in range(len):
@@ -29,70 +18,162 @@ def calcNormalFunc(mean, sd, len):
 
     return f
 
+## Constants ##
 DATASET_PATH = "dataset\\"
+CSV_PATH = "src\\experiments\\algorithms\\data\\"
 BARK_TYPES = 8
 
+## Arrays ##
 avgs_bgr = [[[],[],[]] for x in range(BARK_TYPES)]
+stds_bgr = [[[],[],[]] for x in range(BARK_TYPES)]
 wgts_bgr = [[[],[],[]] for x in range(BARK_TYPES)]
 
 avgs_hsv = [[[],[],[]] for x in range(BARK_TYPES)]
+stds_hsv = [[[],[],[]] for x in range(BARK_TYPES)]
 wgts_hsv = [[[],[],[]] for x in range(BARK_TYPES)]
 
 labels = ["" for x in range(BARK_TYPES)]
-           
+
+## Data extraction ##           
 i = -1
 last_name = ""
 for file in os.listdir(DATASET_PATH):
+    # Store name + iterate tree type #
     name = file.split('_')[0]
     if(name != last_name):
         last_name = name
         i += 1
         labels[i] = name
 
+    # Load picture #
     im_bgr = cv2.imread(os.path.join(DATASET_PATH + file), 1)
     assert im_bgr is not None, "Something went wrong"
     im_hsv = cv2.cvtColor(im_bgr, cv2.COLOR_BGR2HSV)
 
+    # Data storing #
     for j in range(3):
         avgs_bgr[i][j].append(np.mean(im_bgr[:, :, j]))
+        stds_bgr[i][j].append(np.std(im_bgr[:, :, j]))
         wgts_bgr[i][j].append(len(im_bgr[:, :, j]))
 
         avgs_hsv[i][j].append(np.mean(im_hsv[:, :, j]))
+        stds_hsv[i][j].append(np.std(im_hsv[:, :, j]))
         wgts_hsv[i][j].append(len(im_hsv[:, :, j]))
-    
 
+## Arrays ##
 w_avg_bgr = [[0, 0, 0] for x in range(BARK_TYPES)]
 w_std_bgr = [[0, 0, 0] for x in range(BARK_TYPES)]
+w_std_bgr = [[0, 0, 0] for x in range(BARK_TYPES)]
 w_avg_hsv = [[0, 0, 0] for x in range(BARK_TYPES)]
 w_std_hsv = [[0, 0, 0] for x in range(BARK_TYPES)]
+w_std_hsv = [[0, 0, 0] for x in range(BARK_TYPES)]
 
-#TODO: Mag de weighted std berekening met avgs? Denk t nie
+## Statistics calculations ##
 for i in range(BARK_TYPES):
     for j in range(3):
-        w_avg_bgr[i][j] = np.average(avgs_bgr[i][j], weights=wgts_bgr[i][j], axis=0)
+        w_avg_bgr[i][j] = np.average(
-        w_std_bgr[i][j] = calcWeightedStd(avgs_bgr[i][j], wgts_bgr[i][j])
+            avgs_bgr[i][j], 
-
+            weights=wgts_bgr[i][j], 
-        w_avg_hsv[i][j] = np.average(avgs_hsv[i][j], weights=wgts_hsv[i][j], axis=0)
+            axis=0)
-        w_std_hsv[i][j] = calcWeightedStd(avgs_hsv[i][j], wgts_hsv[i][j])
+        w_std_bgr[i][j] = np.average(
+            stds_bgr[i][j], 
+            weights=wgts_bgr[i][j])
 
+        w_avg_hsv[i][j] = np.average(
+            avgs_hsv[i][j], 
+            weights=wgts_hsv[i][j], 
+            axis=0)
+        w_std_hsv[i][j] = np.average(
+            stds_hsv[i][j], 
+            weights=wgts_hsv[i][j])
 
+## Plots + Normal calculations ## 
 fig, axs = plt.subplots(2,4)
 row = 0
 col = -1
 colours = ('b', 'g', 'r')
 for i in range(BARK_TYPES):
+    # Rows and columns # 
     row = 1 if i > 3 else 0
     col += 1
     if i == 4: col = 0
 
+    # Normal and plotting #
     for j in range(3):
         f = calcNormalFunc(w_avg_bgr[i][j], w_std_bgr[i][j], 255)
         label = "$\mu:$ %d, $\sigma: %d$"%(w_avg_bgr[i][j],w_std_bgr[i][j])
         axs[row, col].plot(f, c=colours[j], label=label)
 
+    # Plot params #
     axs[row, col].grid()
     axs[row, col].legend()
     axs[row, col].set_title(str(labels[i]))
-    
+plt.show(block=False)
+
+fig, axs = plt.subplots(2,4)
+row = 0
+col = -1
+for i in range(BARK_TYPES):
+    # Rows and columns # 
+    row = 1 if i > 3 else 0
+    col += 1
+    if i == 4: col = 0
+
+    # Normal and plotting #
+    for j in range(3):
+        f = calcNormalFunc(w_avg_hsv[i][j], w_std_hsv[i][j], 255)
+        label = "$\mu:$ %d, $\sigma: %d$"%(w_avg_hsv[i][j], w_std_hsv[i][j])
+        axs[row, col].plot(f, label=label)
+
+    # Plot params #
+    axs[row, col].grid()
+    axs[row, col].legend()
+    axs[row, col].set_title(str(labels[i]))  
 plt.show()
 
+## CSVs ##
+# BGR mean and std #
+with open(CSV_PATH + "bgr_stats.csv", 'w', newline='') as file:
+    file.write("label,avg_b,avg_g,avg_r,std_b,std_g,std_r,\n")
+    for i in range(BARK_TYPES):
+        file.write(labels[i] + ',')
+        for j in range(3):
+            file.write(str(w_avg_bgr[i][j]) + ',')
+        for j in range(3):
+            file.write(str(w_std_bgr[i][j]) + ',')
+        file.write('\n')
+
+# BGR normal curve per channel #
+id = ('B', 'G', 'R')
+with open(CSV_PATH + "bgr_curve.csv", 'w', newline='') as file:
+    for i in range(BARK_TYPES):
+        for j in range(3):
+            file.write(labels[i] + ',')
+            file.write(id[j] + ',')
+            f = calcNormalFunc(w_avg_bgr[i][j], w_std_bgr[i][j], 255)
+            for k in range(255):
+                file.write(str(f[k]) + ',')
+            file.write('\n')
+
+# HSV mean and std #
+with open(CSV_PATH + "hsv_stats.csv", 'w', newline='') as file:
+    file.write("label,avg_b,avg_g,avg_r,std_b,std_g,std_r,\n")
+    for i in range(BARK_TYPES):
+        file.write(labels[i] + ',')
+        for j in range(3):
+            file.write(str(w_avg_hsv[i][j]) + ',')
+        for j in range(3):
+            file.write(str(w_std_hsv[i][j]) + ',')
+        file.write('\n')
+
+# BGR normal curve per channel #
+id = ('H', 'S', 'V')
+with open(CSV_PATH + "hsv_curve.csv", 'w', newline='') as file:
+    for i in range(BARK_TYPES):
+        for j in range(3):
+            file.write(labels[i] + ',')
+            file.write(id[j] + ',')
+            f = calcNormalFunc(w_avg_hsv[i][j], w_std_hsv[i][j], 255)
+            for k in range(255):
+                file.write(str(f[k]) + ',')
+            file.write('\n')

src/experiments/algorithms/sift_v3.py

@@ -79,7 +79,7 @@ for folder in os.listdir(DATASET_PATH):
 print("Done!")
 
 ## Pandas ##
-with open(CSV_PATH + "data.csv" , 'w', newline='') as file:
+with open(CSV_PATH + "sift.csv" , 'w', newline='') as file:
     for i in range(len(labels)):
         file.write(labels[i] + '\n')
         for j in range(len(tot_magnitudes[i])):