|
|
import cv2 |
|
|
import numpy as np |
|
|
from PIL import Image |
|
|
from backend import config |
|
|
from backend.utils import get_roi |
|
|
from backend.model_handler import model_handler |
|
|
|
|
|
def detect_straight_lines(roi_img): |
|
|
"""Enhanced edge detection focusing on straight lines.""" |
|
|
gray = cv2.cvtColor(roi_img, cv2.COLOR_BGR2GRAY) |
|
|
clahe = cv2.createCLAHE( |
|
|
clipLimit=config.DETECTION_PARAMS['clahe_clip_limit'], |
|
|
tileGridSize=config.DETECTION_PARAMS['clahe_grid_size'] |
|
|
) |
|
|
enhanced = clahe.apply(gray) |
|
|
blurred = cv2.GaussianBlur( |
|
|
enhanced, |
|
|
config.DETECTION_PARAMS['gaussian_kernel'], |
|
|
config.DETECTION_PARAMS['gaussian_sigma'] |
|
|
) |
|
|
edges = cv2.Canny( |
|
|
blurred, |
|
|
config.DETECTION_PARAMS['canny_low'], |
|
|
config.DETECTION_PARAMS['canny_high'] |
|
|
) |
|
|
line_mask = np.zeros_like(edges) |
|
|
lines = cv2.HoughLinesP( |
|
|
edges, |
|
|
rho=1, |
|
|
theta=np.pi/180, |
|
|
threshold=config.DETECTION_PARAMS['hough_threshold'], |
|
|
minLineLength=config.DETECTION_PARAMS['min_line_length'], |
|
|
maxLineGap=config.DETECTION_PARAMS['max_line_gap'] |
|
|
) |
|
|
if lines is not None: |
|
|
for line in lines: |
|
|
x1, y1, x2, y2 = line[0] |
|
|
cv2.line(line_mask, (x1, y1), (x2, y2), 255, 2) |
|
|
return line_mask |
|
|
|
|
|
def simple_edge_detection(roi_img): |
|
|
"""Simple edge detection.""" |
|
|
gray = cv2.cvtColor(roi_img, cv2.COLOR_BGR2GRAY) |
|
|
return cv2.Canny(gray, 50, 150) |
|
|
|
|
|
def ribbon(image_path): |
|
|
"""Detect the presence of a ribbon in an image.""" |
|
|
image = cv2.imread(image_path) |
|
|
if image is None: |
|
|
raise ValueError(f"Could not read image: {image_path}") |
|
|
|
|
|
h, w = image.shape[:2] |
|
|
edge_present = [] |
|
|
|
|
|
for i, roi in enumerate(config.ROIS): |
|
|
x1, y1, x2, y2 = [int(coord * (w if i % 2 == 0 else h)) for i, coord in enumerate(roi)] |
|
|
roi_img = image[y1:y2, x1:x2] |
|
|
|
|
|
if i < 6: |
|
|
edges = detect_straight_lines(roi_img) |
|
|
edge_present.append(np.sum(edges) > edges.size * config.DETECTION_PARAMS['edge_pixel_threshold']) |
|
|
else: |
|
|
edges = simple_edge_detection(roi_img) |
|
|
edge_present.append(np.any(edges)) |
|
|
|
|
|
result = all(edge_present[:6]) and not edge_present[6] and not edge_present[7] and not edge_present[8] |
|
|
return {"No Ribbon": 0 if result else 1} |
|
|
|
|
|
def image_quality(image_path): |
|
|
""" |
|
|
Check if an image is low resolution or poor quality. |
|
|
""" |
|
|
try: |
|
|
image = Image.open(image_path) |
|
|
width, height = image.size |
|
|
pixel_count = width * height |
|
|
|
|
|
if width < config.MIN_WIDTH or height < config.MIN_HEIGHT or pixel_count < config.MIN_PIXEL_COUNT: |
|
|
return {"Bad Image Quality": 1} |
|
|
|
|
|
grayscale_image = image.convert("L") |
|
|
pixel_array = np.array(grayscale_image) |
|
|
variance = np.var(pixel_array) |
|
|
|
|
|
if variance < config.PIXEL_VARIANCE_THRESHOLD: |
|
|
return {"Bad Image Quality": 1} |
|
|
|
|
|
return {"Bad Image Quality": 0} |
|
|
|
|
|
except Exception as e: |
|
|
print(f"Error processing image: {e}") |
|
|
return {"Bad Image Quality": 1} |
|
|
|
|
|
def gnc(image_path): |
|
|
"""Check for gestures/coach marks and display the image.""" |
|
|
image = get_roi(image_path, *config.GNC) |
|
|
gnc_text = model_handler.intern(image, config.PGNC, 900).lower() |
|
|
|
|
|
return {"Visual Gesture or Icon": 1 if 'yes' in gnc_text else 0} |
|
|
|
