File size: 2,192 Bytes
4c1ab88 |
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 |
import cv2 as cv
import numpy as np
from scipy.signal import find_peaks
from PIL import Image # Import PIL
class ObstructionDetector:
def __init__(self, threshold=500):
self.threshold = threshold
def preprocess_image(self, image):
# Convert the image to grayscale if it's a color image
if len(image.shape) == 3:
image = cv.cvtColor(image, cv.COLOR_BGR2GRAY)
# Apply Gaussian blur to reduce noise
preprocessed_image = cv.GaussianBlur(image, (5, 5), 0)
# Perform other preprocessing steps as needed (e.g., contrast adjustment, histogram equalization)
return preprocessed_image
def plot_histogram(self, image):
# Calculate the histogram
histogram = cv.calcHist([image], [0], None, [256], [0, 256])
# Smoothing the histogram using a simple moving average (window size = 5)
kernel = np.ones((5, 1)) / 5
smoothed_histogram = cv.filter2D(histogram, -1, kernel)
return smoothed_histogram
def count_histogram_peaks(self, smoothed_histogram):
# Find peaks in the smoothed histogram with frequency greater than the threshold
peaks, _ = find_peaks(smoothed_histogram.flatten(), height=self.threshold)
return peaks
def detect_obstruction(self, pil_image): # Accept PIL image directly
# Convert PIL image to NumPy array
img = np.array(pil_image)
# Preprocess the image
preprocessed_img = self.preprocess_image(img)
# Count the number of peaks in the smoothed histogram above the threshold
smoothed_histogram = self.plot_histogram(preprocessed_img)
peaks = self.count_histogram_peaks(smoothed_histogram)
# Check if peaks are too close together
peak_spacing = np.diff(peaks)
if len(peak_spacing) == 0 or np.all(peak_spacing < 10):
report = "A imagem NÃO contém obstrução significativa | e NÃO possui múltiplas distribuições de densidade claramente distintas."
else:
report = "A imagem contém obstrução significativa | possui múltiplas distribuições de densidade claramente distintas."
return report |