Spaces:
Sleeping
Sleeping
File size: 19,582 Bytes
fdcec08 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 |
import cv2
import numpy as np
import pandas as pd
from functools import reduce
def detect_graph_boundaries(img):
height, width = img.shape[:2]
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
_, thresh = cv2.threshold(gray, 200, 255, cv2.THRESH_BINARY_INV)
col_sums = np.sum(thresh, axis=0) / 255
is_line = col_sums > (height * 0.40)
line_indices = np.where(is_line)[0]
start_x = 0
if len(line_indices) > 0:
left_lines = [x for x in line_indices if x < width * 0.2 and x > 5]
if left_lines: start_x = left_lines[0]
end_x = width - 1
if len(line_indices) > 0:
right_margin = width * 0.95
right_lines = [x for x in line_indices if x > right_margin]
if right_lines: end_x = right_lines[-1]
# Create debug image
debug_img = img.copy()
cv2.line(debug_img, (int(start_x), 0), (int(start_x), height), (0, 255, 0), 3)
cv2.line(debug_img, (int(end_x), 0), (int(end_x), height), (0, 0, 255), 3)
return int(start_x), int(end_x), debug_img
def extract_line_mask(img_cropped, line_color, saturation_factor, gap_fill_size, noise_threshold):
# Boost Saturation
hsv_pre = cv2.cvtColor(img_cropped, cv2.COLOR_BGR2HSV)
h, s, v = cv2.split(hsv_pre)
s = np.clip(s.astype(np.float32) * saturation_factor, 0, 255).astype(np.uint8)
hsv = cv2.merge((h, s, v))
b, g, r = cv2.split(cv2.cvtColor(hsv, cv2.COLOR_HSV2BGR))
mask = None
if line_color == "Green":
lower = np.array([35, 20, 100]); upper = np.array([75, 255, 255])
mask_hsv = cv2.inRange(hsv, lower, upper)
diff_gb = g.astype(np.int16) - b.astype(np.int16)
diff_gr = g.astype(np.int16) - r.astype(np.int16)
mask_channel = np.zeros_like(g, dtype=np.uint8)
mask_channel[(diff_gb > 20) & (diff_gr > 10)] = 255
mask = cv2.bitwise_and(mask_hsv, mask_channel)
elif line_color == "Blue (Cyan)":
lower = np.array([80, 20, 100]); upper = np.array([100, 255, 255])
mask_hsv = cv2.inRange(hsv, lower, upper)
diff_br = b.astype(np.int16) - r.astype(np.int16)
mask_channel = np.zeros_like(b, dtype=np.uint8)
mask_channel[diff_br > 20] = 255
mask = cv2.bitwise_and(mask_hsv, mask_channel)
elif line_color == "Red":
lower1 = np.array([0, 20, 100]); upper1 = np.array([10, 255, 255])
lower2 = np.array([170, 20, 100]); upper2 = np.array([180, 255, 255])
mask_hsv = cv2.bitwise_or(cv2.inRange(hsv, lower1, upper1), cv2.inRange(hsv, lower2, upper2))
diff_rg = r.astype(np.int16) - g.astype(np.int16)
diff_rb = r.astype(np.int16) - b.astype(np.int16)
mask_channel = np.zeros_like(r, dtype=np.uint8)
mask_channel[(diff_rg > 20) & (diff_rb > 20)] = 255
mask = cv2.bitwise_and(mask_hsv, mask_channel)
if mask is None: return None
# Noise/Gap cleanup
contours, _ = cv2.findContours(mask, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
mask_clean = np.zeros_like(mask)
for cnt in contours:
# Reduced noise threshold slightly to keep thin spikes
if cv2.contourArea(cnt) > (noise_threshold * 0.5):
cv2.drawContours(mask_clean, [cnt], -1, 255, -1)
mask = mask_clean
if gap_fill_size > 0:
k_h = np.ones((1, gap_fill_size), np.uint8)
close_h = cv2.morphologyEx(mask, cv2.MORPH_CLOSE, k_h)
k_v = np.ones((gap_fill_size, 1), np.uint8)
close_v = cv2.morphologyEx(mask, cv2.MORPH_CLOSE, k_v)
mask = cv2.bitwise_or(close_h, close_v)
# Replaced the 3x3 CLOSE with a smaller one to prevent merging nearby spikes too aggressively
mask = cv2.morphologyEx(mask, cv2.MORPH_CLOSE, np.ones((2,2), np.uint8))
# --- FIX 1: REMOVED MORPH_OPEN ---
# The previous code had: mask = cv2.morphologyEx(mask, cv2.MORPH_OPEN, np.ones((2,2), np.uint8))
# This erodes the image. If a spike is very sharp (1-2px tip), this line deletes the tip.
# We remove it to preserve high-frequency details.
return mask
def generate_curve_data(mask, name_upper, name_lower):
height, width = mask.shape
data = []
prev_top, prev_bot = None, None
proximity_thresh = 10
for x in range(width):
col = mask[:, x]
indices = np.where(col > 0)[0]
val_top, val_bot = None, None
if len(indices) > 0:
y_min, y_max = indices[0], indices[-1]
graph_y_top = height - y_min
graph_y_bot = height - y_max
# If the line is thin, it's a single curve
if abs(y_max - y_min) <= proximity_thresh:
current_val = height - int((y_min + y_max) / 2)
# Simple tracking to decide if it belongs to top or bottom curve if they were split previously
if prev_top is None and prev_bot is None: val_top = current_val
elif prev_top is not None and prev_bot is None: val_top = current_val
elif prev_top is None and prev_bot is not None: val_bot = current_val
else:
if abs(current_val - prev_top) <= abs(current_val - prev_bot): val_top = current_val
else: val_bot = current_val
else:
# Vertical line (spike) or filled area
val_top = graph_y_top
val_bot = graph_y_bot
if val_top is not None: prev_top = val_top
if val_bot is not None: prev_bot = val_bot
data.append({"X": x, name_upper: val_top, name_lower: val_bot})
df = pd.DataFrame(data)
# Using 'pchip' or 'linear' interpolation.
# 'linear' is safer for sharp spikes. 'pchip' can overshoot.
df[name_upper] = df[name_upper].interpolate(method='linear', limit_direction='both')
df[name_lower] = df[name_lower].interpolate(method='linear', limit_direction='both')
return df
def process_uploaded_image(file_bytes, sat_factor, gap_size, noise_threshold, crop_enabled, total_duration):
# 1. Decode Image
file_bytes = np.asarray(bytearray(file_bytes), dtype=np.uint8)
img_orig = cv2.imdecode(file_bytes, 1)
# 2. Crop
debug_img_bounds = img_orig.copy()
sx, ex = 0, img_orig.shape[1]
if crop_enabled:
sx, ex, debug_img_bounds = detect_graph_boundaries(img_orig)
img_working = img_orig[:, sx:ex]
else:
img_working = img_orig
if img_working.shape[1] == 0:
return None, None, None, "Crop failed."
# 3. Process Colors
configs = [
("Red", "Red", ("Travel", "C1")),
("Green", "Green", ("Resistance", "C2")),
("Blue (Cyan)", "Blue", ("Current", "C3"))
]
dfs = []
debug_images = {}
debug_images["Boundaries"] = debug_img_bounds
height, width = img_working.shape[:2]
for color_key, _, col_names in configs:
mask = extract_line_mask(img_working, color_key, sat_factor, gap_size, noise_threshold)
if mask is not None:
colored_mask = np.zeros_like(img_working)
colored_mask[mask > 0] = [0, 255, 0]
overlay = cv2.addWeighted(img_working, 0.7, colored_mask, 0.3, 0)
debug_images[color_key] = cv2.cvtColor(overlay, cv2.COLOR_BGR2RGB)
df_curve = generate_curve_data(mask, col_names[0], col_names[1])
dfs.append(df_curve)
else:
df_empty = pd.DataFrame({"X": range(width), col_names[0]: np.nan, col_names[1]: np.nan})
dfs.append(df_empty)
# 4. Merge
if dfs:
final_df = reduce(lambda left, right: pd.merge(left, right, on='X', how='outer'), dfs)
# STEP A: GENERATE TIME COLUMN
cols = ['X', 'Travel', 'C1', 'Resistance', 'C2', 'Current', 'C3']
existing_cols = [c for c in cols if c in final_df.columns]
if 'X' in final_df.columns:
time_per_pixel = total_duration / width
final_df['Time (ms)'] = final_df['X'] * time_per_pixel
existing_cols.insert(1, 'Time (ms)')
else:
return None, None, None, "X-axis alignment failed."
# STEP B: CALCULATE BASELINES & CLEANUP
for col in ['Current', 'Travel']:
if col in final_df.columns:
# Baseline - keep this simple to detect zero offset
baseline_val = 0
start_window = final_df[final_df['Time (ms)'] <= 30]
if not start_window.empty and start_window[col].notna().any():
baseline_val = start_window[col].mean()
else:
valid_idx = final_df[col].first_valid_index()
if valid_idx is not None: baseline_val = final_df.loc[valid_idx, col]
# --- FIX 2: Relaxed Baseline Clipping ---
# Previous code strictly deleted anything below baseline.
# Spikes often oscillate. We allow small dips now.
end_x = width - 1
if len(line_indices) > 0:
right_margin = width * 0.95
right_lines = [x for x in line_indices if x > right_margin]
if right_lines: end_x = right_lines[-1]
# Create debug image
debug_img = img.copy()
cv2.line(debug_img, (int(start_x), 0), (int(start_x), height), (0, 255, 0), 3)
cv2.line(debug_img, (int(end_x), 0), (int(end_x), height), (0, 0, 255), 3)
return int(start_x), int(end_x), debug_img
def extract_line_mask(img_cropped, line_color, saturation_factor, gap_fill_size, noise_threshold):
# Boost Saturation
hsv_pre = cv2.cvtColor(img_cropped, cv2.COLOR_BGR2HSV)
h, s, v = cv2.split(hsv_pre)
s = np.clip(s.astype(np.float32) * saturation_factor, 0, 255).astype(np.uint8)
hsv = cv2.merge((h, s, v))
b, g, r = cv2.split(cv2.cvtColor(hsv, cv2.COLOR_HSV2BGR))
mask = None
if line_color == "Green":
lower = np.array([35, 20, 100]); upper = np.array([75, 255, 255])
mask_hsv = cv2.inRange(hsv, lower, upper)
diff_gb = g.astype(np.int16) - b.astype(np.int16)
diff_gr = g.astype(np.int16) - r.astype(np.int16)
mask_channel = np.zeros_like(g, dtype=np.uint8)
mask_channel[(diff_gb > 20) & (diff_gr > 10)] = 255
mask = cv2.bitwise_and(mask_hsv, mask_channel)
elif line_color == "Blue (Cyan)":
lower = np.array([80, 20, 100]); upper = np.array([100, 255, 255])
mask_hsv = cv2.inRange(hsv, lower, upper)
diff_br = b.astype(np.int16) - r.astype(np.int16)
mask_channel = np.zeros_like(b, dtype=np.uint8)
mask_channel[diff_br > 20] = 255
mask = cv2.bitwise_and(mask_hsv, mask_channel)
elif line_color == "Red":
lower1 = np.array([0, 20, 100]); upper1 = np.array([10, 255, 255])
lower2 = np.array([170, 20, 100]); upper2 = np.array([180, 255, 255])
mask_hsv = cv2.bitwise_or(cv2.inRange(hsv, lower1, upper1), cv2.inRange(hsv, lower2, upper2))
diff_rg = r.astype(np.int16) - g.astype(np.int16)
diff_rb = r.astype(np.int16) - b.astype(np.int16)
mask_channel = np.zeros_like(r, dtype=np.uint8)
mask_channel[(diff_rg > 20) & (diff_rb > 20)] = 255
mask = cv2.bitwise_and(mask_hsv, mask_channel)
if mask is None: return None
# Noise/Gap cleanup
contours, _ = cv2.findContours(mask, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
mask_clean = np.zeros_like(mask)
for cnt in contours:
# Reduced noise threshold slightly to keep thin spikes
if cv2.contourArea(cnt) > (noise_threshold * 0.5):
cv2.drawContours(mask_clean, [cnt], -1, 255, -1)
mask = mask_clean
if gap_fill_size > 0:
k_h = np.ones((1, gap_fill_size), np.uint8)
close_h = cv2.morphologyEx(mask, cv2.MORPH_CLOSE, k_h)
k_v = np.ones((gap_fill_size, 1), np.uint8)
close_v = cv2.morphologyEx(mask, cv2.MORPH_CLOSE, k_v)
mask = cv2.bitwise_or(close_h, close_v)
# Replaced the 3x3 CLOSE with a smaller one to prevent merging nearby spikes too aggressively
mask = cv2.morphologyEx(mask, cv2.MORPH_CLOSE, np.ones((2,2), np.uint8))
# --- FIX 1: REMOVED MORPH_OPEN ---
# The previous code had: mask = cv2.morphologyEx(mask, cv2.MORPH_OPEN, np.ones((2,2), np.uint8))
# This erodes the image. If a spike is very sharp (1-2px tip), this line deletes the tip.
# We remove it to preserve high-frequency details.
return mask
def generate_curve_data(mask, name_upper, name_lower):
height, width = mask.shape
data = []
prev_top, prev_bot = None, None
proximity_thresh = 10
for x in range(width):
col = mask[:, x]
indices = np.where(col > 0)[0]
val_top, val_bot = None, None
if len(indices) > 0:
y_min, y_max = indices[0], indices[-1]
graph_y_top = height - y_min
graph_y_bot = height - y_max
# If the line is thin, it's a single curve
if abs(y_max - y_min) <= proximity_thresh:
current_val = height - int((y_min + y_max) / 2)
# Simple tracking to decide if it belongs to top or bottom curve if they were split previously
if prev_top is None and prev_bot is None: val_top = current_val
elif prev_top is not None and prev_bot is None: val_top = current_val
elif prev_top is None and prev_bot is not None: val_bot = current_val
else:
if abs(current_val - prev_top) <= abs(current_val - prev_bot): val_top = current_val
else: val_bot = current_val
else:
# Vertical line (spike) or filled area
val_top = graph_y_top
val_bot = graph_y_bot
if val_top is not None: prev_top = val_top
if val_bot is not None: prev_bot = val_bot
data.append({"X": x, name_upper: val_top, name_lower: val_bot})
df = pd.DataFrame(data)
# Using 'pchip' or 'linear' interpolation.
# 'linear' is safer for sharp spikes. 'pchip' can overshoot.
df[name_upper] = df[name_upper].interpolate(method='linear', limit_direction='both')
df[name_lower] = df[name_lower].interpolate(method='linear', limit_direction='both')
return df
def process_uploaded_image(file_bytes, sat_factor, gap_size, noise_threshold, crop_enabled, total_duration):
# 1. Decode Image
file_bytes = np.asarray(bytearray(file_bytes), dtype=np.uint8)
img_orig = cv2.imdecode(file_bytes, 1)
# 2. Crop
debug_img_bounds = img_orig.copy()
sx, ex = 0, img_orig.shape[1]
if crop_enabled:
sx, ex, debug_img_bounds = detect_graph_boundaries(img_orig)
img_working = img_orig[:, sx:ex]
else:
img_working = img_orig
if img_working.shape[1] == 0:
return None, None, None, "Crop failed."
# 3. Process Colors
configs = [
("Red", "Red", ("Travel", "C1")),
("Green", "Green", ("Resistance", "C2")),
("Blue (Cyan)", "Blue", ("Current", "C3"))
]
dfs = []
debug_images = {}
debug_images["Boundaries"] = debug_img_bounds
height, width = img_working.shape[:2]
for color_key, _, col_names in configs:
mask = extract_line_mask(img_working, color_key, sat_factor, gap_size, noise_threshold)
if mask is not None:
colored_mask = np.zeros_like(img_working)
colored_mask[mask > 0] = [0, 255, 0]
overlay = cv2.addWeighted(img_working, 0.7, colored_mask, 0.3, 0)
debug_images[color_key] = cv2.cvtColor(overlay, cv2.COLOR_BGR2RGB)
df_curve = generate_curve_data(mask, col_names[0], col_names[1])
dfs.append(df_curve)
else:
df_empty = pd.DataFrame({"X": range(width), col_names[0]: np.nan, col_names[1]: np.nan})
dfs.append(df_empty)
# 4. Merge
if dfs:
final_df = reduce(lambda left, right: pd.merge(left, right, on='X', how='outer'), dfs)
# STEP A: GENERATE TIME COLUMN
cols = ['X', 'Travel', 'C1', 'Resistance', 'C2', 'Current', 'C3']
existing_cols = [c for c in cols if c in final_df.columns]
if 'X' in final_df.columns:
time_per_pixel = total_duration / width
final_df['Time (ms)'] = final_df['X'] * time_per_pixel
existing_cols.insert(1, 'Time (ms)')
else:
return None, None, None, "X-axis alignment failed."
# STEP B: CALCULATE BASELINES & CLEANUP
for col in ['Current', 'Travel']:
if col in final_df.columns:
# Baseline - keep this simple to detect zero offset
baseline_val = 0
start_window = final_df[final_df['Time (ms)'] <= 30]
if not start_window.empty and start_window[col].notna().any():
baseline_val = start_window[col].mean()
else:
valid_idx = final_df[col].first_valid_index()
if valid_idx is not None: baseline_val = final_df.loc[valid_idx, col]
# --- FIX 2: Relaxed Baseline Clipping ---
# Previous code strictly deleted anything below baseline.
# Spikes often oscillate. We allow small dips now.
# Only clip if it is essentially noise below 0 (assuming values are positive)
# If your graph allows negative values, remove this line entirely.
if baseline_val > 0:
final_df.loc[final_df[col] < (baseline_val * 0.8), col] = np.nan
final_df[col] = final_df[col].interpolate(method='linear', limit_direction='both')
# --- FIX 3: REMOVED "Middle Section Cleanup" ---
# The previous code deleted data between 120ms and 250ms if it was near the baseline.
# This was the primary cause of spikes disappearing in that region.
# The code block is deleted here.
# STEP C: AUXILIARY CURVE LOGIC
# Purpose: Clean auxiliary curves (C1, C2, C3) that represent the bottom edge of thick lines
# The auxiliary curve should never be ABOVE the main curve at the same X position
pairs = [('C1', 'Travel'), ('C2', 'Resistance'), ('C3', 'Current')]
for lower, upper in pairs:
if lower in final_df.columns and upper in final_df.columns:
if not final_df[upper].isnull().all():
# FIX: Compare at same X position to preserve spikes
# Previous logic: final_df[lower] > min_upper (global minimum)
# - This deleted spike data because spike bottoms exceeded the global min
# New logic: final_df[lower] > final_df[upper] (position-wise comparison)
# - This only deletes truly invalid crossovers at the same X
# - Preserves spikes where upper and lower legitimately differ
invalid_mask = final_df[lower] > final_df[upper]
final_df.loc[invalid_mask, lower] = np.nan
# Final global cleanup
for col in ['Current', 'Travel', 'C1', 'C2', 'C3']:
if col in final_df.columns:
final_df[col] = final_df[col].interpolate(method='linear', limit_direction='both')
return final_df[existing_cols], debug_images, (sx, ex), None
return None, None, None, "No data extracted." |