plot_model plots now the previously fitted model and does not fit a new model

Author: O-Sven

ledsa/analysis/plot_functions.py

@@ -14,6 +14,7 @@
 sep = os.path.sep
 
 # dictionary of the fit parameter positions
+# only used as reference
 par_dic = {
     "x": 2,
     "y": 3,
@@ -85,7 +86,8 @@ def plot_t_fitpar(fig, led_id, fit_par, channel, image_id_start, image_id_finish
 def plot_t_fitpar_with_moving_average(fig, led_id, fit_par, channel, image_id_start, image_id_finish, box_size=61):
     """Plots the time development of a fit parameter and its moving average"""
     plot_info = _calc_t_fitpar_plot_info(led_id, fit_par, channel, image_id_start, image_id_finish)
-    average = plot_info[fit_par].rolling(box_size, center=True, win_type='gaussian').sum(std=10) / 10 / np.sqrt(2 * np.pi)
+    average = plot_info[fit_par].rolling(box_size, center=True, win_type='gaussian').sum(std=10) / (10 *
+                                                                                                    np.sqrt(2 * np.pi))
 
     ax = fig.gca(xlabel='time[s]', ylabel=fit_par)
     plot, = ax.plot(plot_info['experiment_time'], plot_info[fit_par], alpha=0.2)
@@ -119,7 +121,6 @@ def plot_led_with_fit(channel, time, led_id, window_radius=10):
 
     plt.title(f'Channel {channel}, Image {img_id}, LED {led_id}')
 
-    # plt.savefig(model.png)
     plt.show()
 
 
@@ -139,11 +140,13 @@ def plot_model(fig, channel, img_id, led_id, window_radius):
     mesh = np.meshgrid(np.linspace(0.5, 2 * window_radius - 0.5, 2 * window_radius),
                        np.linspace(0.5, 2 * window_radius - 0.5, 2 * window_radius))
 
-    fit_results = fit_led(img_id, led_id, channel)
-    model_params = fit_results.x
-    print(model_params)
-    print(fit_results.keys())
-    print(fit_results)
+    # fit model
+    # fit_results = fit_led(img_id, led_id, channel)
+    # model_params = fit_results.x
+
+    # load model
+    model_params = load_model(img_id, led_id, channel, window_radius)
+
     led_model = led.led_fit(mesh[0], mesh[1], model_params[0], model_params[1], model_params[2], model_params[3],
                             model_params[4], model_params[5], model_params[6], model_params[7])
 
@@ -154,8 +157,8 @@ def plot_model(fig, channel, img_id, led_id, window_radius):
     con = ax.contour(mesh[0], mesh[1], led_model, levels=10, alpha=0.9)
     fig.colorbar(mappable=con, ax=ax)
     ax.scatter(model_params[0], model_params[1], color='Red')
-    plt.text(0, window_radius * 2.2, f'Num. of Iterations: {fit_results.nit} -/- l2 + penalty: {fit_results.fun:.4}',
-             ha='left')
+    # plt.text(0, window_radius * 2.2, f'Num. of Iterations: {fit_results.nit} -/- l2 + penalty: {fit_results.fun:.4}',
+    #          ha='left')
 
     plt.figure(current_fig.number)
 
@@ -222,17 +225,22 @@ def fit_led(img_id, led_id, channel):
     return fit_res
 
 
+def load_model(img_id, led_id, channel, window_radius=10):
+    fit_parameters = calc.read_hdf(channel)
+    model_params = np.array(fit_parameters.loc[img_id, led_id])[1:9]
+    pix_pos = get_led_pos(led_id)
+    model_params[0:2] = model_params[0:2] - pix_pos + window_radius
+    return model_params
+
+
 def get_led_img(time, led_id, window_radius=10):
     filename = led.get_img_name(led.get_img_id_from_time(time))
-    print(filename)
     path = get_img_path()
     led_im = Image.open(path + filename)
 
     x, y = get_led_pos(led_id)
-    print(x, y)
     led_im = led_im.crop((y - window_radius,
                           x - window_radius,
                           y + window_radius,
                           x + window_radius))
     return led_im
-

ledsa/core/_led_helper.py

@@ -82,8 +82,8 @@ def get_img_data(config, build_experiment_infos=False, build_analysis_infos=Fals
                     experiment_time = experiment_time.total_seconds()
                     time_diff = config[build_type]['exif_time_infront_real_time'].split('.')
                     time = date_time_img - timedelta(seconds=int(time_diff[0]), milliseconds=int(time_diff[1]))
-                    img_data += (str(img_idx) + ',' + config[build_type]['img_name_string'].format(int(img_number)) + ',' +
-                                 time.strftime('%H:%M:%S') + ',' + str(experiment_time) + '\n')
+                    img_data += (str(img_idx) + ',' + config[build_type]['img_name_string'].format(int(img_number)) +
+                                 ',' + time.strftime('%H:%M:%S') + ',' + str(experiment_time) + '\n')
                     img_idx += 1
     return img_data
 

ledsa/tests/log_analysis.py

@@ -48,7 +48,7 @@ def plot_image(self):
         mesh = np.meshgrid(np.linspace(0.5, self.nx - 0.5, self.nx), np.linspace(0.5, self.ny - 0.5, self.ny))
 
         led_model = led.led_fit(mesh[0], mesh[1], self.fit[0], self.fit[1], self.fit[2], self.fit[3], self.fit[4],
-                            self.fit[5], self.fit[6], self.fit[7])
+                                self.fit[5], self.fit[6], self.fit[7])
 
         fig, ax = plt.subplots(1, 2, dpi=600)
 
@@ -59,7 +59,7 @@ def plot_image(self):
         ax[1].imshow(led_model, cmap='Greys')
 
         ampl = np.max(np.abs(data[s] - led_model))  # 0.25
-        maxA = 255 # np.max(np.abs(data[s]))
+        maxA = 255  # np.max(np.abs(data[s]))
 
         im2 = ax[1].imshow((data[s] - led_model)/maxA, cmap='seismic', vmin=-ampl/maxA, vmax=ampl/maxA)
         plt.colorbar(mappable=im2)