顏色、形狀和紋理：使用 OpenCV 進(jìn)行特征提取
點(diǎn)擊上方“小白學(xué)視覺”，選擇加"星標(biāo)"或“置頂” 重磅干貨，第一時間送達(dá) 
圖像處理所做的只是從圖像中提取有用的信息，從而減少數(shù)據(jù)量，但保留描述圖像特征的像素。
1. 顏色
-
色調(diào)：描述主波長，是指定顏色的通道 -
飽和度：描述色調(diào)/顏色的純度/色調(diào) -
值：描述顏色的強(qiáng)度
import cv2
from google.colab.patches import cv2_imshow
image = cv2.imread(image_file)
hsv_image = cv2.cvtColor(image, cv2.COLOR_BGR2HSV)
cv2_imshow(hsv_image)
-
L：描述顏色的亮度，與強(qiáng)度互換使用 -
A : 顏色成分范圍，從綠色到品紅色 -
B：從藍(lán)色到黃色的顏色分量
import cv2
from google.colab.patches import cv2_imshow
image = cv2.imread(image_file)
lab_image = cv2.cvtColor(image, cv2.COLOR_BGR2LAB)
cv2_imshow(lab_image)
-
Y : 伽馬校正后從 RGB 顏色空間獲得的亮度 -
Cr：描述紅色 (R) 分量與亮度的距離 -
Cb：描述藍(lán)色 (B) 分量與亮度的距離
import cv2
from google.colab.patches import cv2_imshow
image = cv2.imread(image_file)
ycrcb_image = cv2.cvtColor(image, cv2.COLOR_BGR2YCrCb)
cv2_imshow(ycrcb_image)
import cv2
from google.colab.patches import cv2_imshow
import numpy as np
import plotly.figure_factory as ff
# Check the distribution of red values
red_values = []
for i in range(len(images)):
red_value = np.mean(images[i][:, :, 0])
red_values.append(red_value)
# Check the distribution of green values
green_values = []
for i in range(len(images)):
green_value = np.mean(images[i][:, :, 1])
green_values.append(green_value)
# Check the distribution of blue values
blue_values = []
for i in range(len(images)):
blue_value = np.mean(images[i][:, :, 2])
blue_values.append(blue_value)
# Plotting the histogram
fig = ff.create_distplot([red_values, green_values, blue_values], group_labels=["R", "G", "B"], colors=['red', 'green', 'blue'])
fig.update_layout(showlegend=True, template="simple_white")
fig.update_layout(title_text='Distribution of channel values across images in RGB')
fig.data[0].marker.line.color = 'rgb(0, 0, 0)'
fig.data[0].marker.line.width = 0.5
fig.data[1].marker.line.color = 'rgb(0, 0, 0)'
fig.data[1].marker.line.width = 0.5
fig.data[2].marker.line.color = 'rgb(0, 0, 0)'
fig.data[2].marker.line.width = 0.5
fig
import cv2
from google.colab.patches import cv2_imshow
# Reading the original image
image_spot = cv2.imread(image_file)
cv2_imshow(image_spot)
# Converting it to HSV color space
hsv_image_spot = cv2.cvtColor(image_spot, cv2.COLOR_BGR2HSV)
cv2_imshow(hsv_image_spot)
# Setting the black pixel mask and perform bitwise_and to get only the black pixels
mask = cv2.inRange(hsv_image_spot, (0, 0, 0), (180, 255, 40))
masked = cv2.bitwise_and(hsv_image_spot, hsv_image_spot, mask=mask)
cv2_imshow(masked)
import cv2
from google.colab.patches import cv2_imshow
image_spot_reshaped = image_spot.reshape((image_spot.shape[0] * image_spot.shape[1], 3))
# convert to np.float32
Z = np.float32(image_spot_reshaped)
# define criteria, number of clusters(K) and apply kmeans()
criteria = (cv2.TERM_CRITERIA_EPS + cv2.TERM_CRITERIA_MAX_ITER, 10, 1.0)
K = 2
ret, label, center = cv2.kmeans(Z, K, None, criteria, 10, cv2.KMEANS_RANDOM_CENTERS)
# Now convert back into uint8, and make original image
center = np.uint8(center)
res = center[label.flatten()]
res2 = res.reshape((image_spot.shape))
cv2_imshow(res2)
2. 形狀
-
矩 -
輪廓面積 -
輪廓周長 -
輪廓近似 -
凸包 -
凸性檢測 -
矩形邊界 -
最小外接圓 -
擬合橢圓 -
擬合直線
3. 紋理
-
對比度：測量灰度共生矩陣的局部變化。 -
相關(guān)性：測量指定像素對的聯(lián)合概率出現(xiàn)。 -
平方：提供 GLCM 中元素的平方和。也稱為均勻性或角二階矩。 -
同質(zhì)性：測量 GLCM 中元素分布與 GLCM 對角線的接近程度。
import cv2
from google.colab.patches import cv2_imshow
image_spot = cv2.imread(image_file)
gray = cv2.cvtColor(image_spot, cv2.COLOR_BGR2GRAY)
# Find the GLCM
import skimage.feature as feature
# Param:
# source image
# List of pixel pair distance offsets - here 1 in each direction
# List of pixel pair angles in radians
graycom = feature.greycomatrix(gray, [1], [0, np.pi/4, np.pi/2, 3*np.pi/4], levels=256)
# Find the GLCM properties
contrast = feature.greycoprops(graycom, 'contrast')
dissimilarity = feature.greycoprops(graycom, 'dissimilarity')
homogeneity = feature.greycoprops(graycom, 'homogeneity')
energy = feature.greycoprops(graycom, 'energy')
correlation = feature.greycoprops(graycom, 'correlation')
ASM = feature.greycoprops(graycom, 'ASM')
print("Contrast: {}".format(contrast))
print("Dissimilarity: {}".format(dissimilarity))
print("Homogeneity: {}".format(homogeneity))
print("Energy: {}".format(energy))
print("Correlation: {}".format(correlation))
print("ASM: {}".format(ASM))
-
https://www.pyimagesearch.com/2015/12/07/local-binary-patterns-with-python-opencv/ -
https://towardsdatascience.com/face-recognition-how-lbph-works-90ec258c3d6b
import cv2
from google.colab.patches import cv2_imshow
class LocalBinaryPatterns:
def __init__(self, numPoints, radius):
self.numPoints = numPoints
self.radius = radius
def describe(self, image, eps = 1e-7):
lbp = feature.local_binary_pattern(image, self.numPoints, self.radius, method="uniform")
(hist, _) = np.histogram(lbp.ravel(), bins=np.arange(0, self.numPoints+3), range=(0, self.numPoints + 2))
# Normalize the histogram
hist = hist.astype('float')
hist /= (hist.sum() + eps)
return hist, lbp
image = cv2.imread(image_file)
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
desc = LocalBinaryPatterns(24, 8)
hist, lbp = desc.describe(gray)
print("Histogram of Local Binary Pattern value: {}".format(hist))
contrast = contrast.flatten()
dissimilarity = dissimilarity.flatten()
homogeneity = homogeneity.flatten()
energy = energy.flatten()
correlation = correlation.flatten()
ASM = ASM.flatten()
hist = hist.flatten()
features = np.concatenate((contrast, dissimilarity, homogeneity, energy, correlation, ASM, hist), axis=0)
cv2_imshow(gray)
cv2_imshow(lbp)
結(jié)論
參考
-
https://www.analyticsvidhya.com/blog/2019/08/3-techniques-extract-features-from-image-data-machine-learning-python/ -
https://www.mygreatlearning.com/blog/feature-extraction-in-image-processing/ -
https://thevatsalsaglani.medium.com/feature-extraction-from-medical-images-and-an-introduction-to-xtract-features-9a225243e94b -
http://www.scholarpedia.org/article/Local_Binary_Patterns -
https://medium.com/mlearning-ai/mlearning-ai-submission-suggestions-b51e2b130bfb
下載1：OpenCV-Contrib擴(kuò)展模塊中文版教程
在「小白學(xué)視覺」公眾號后臺回復(fù)：擴(kuò)展模塊中文教程，即可下載全網(wǎng)第一份OpenCV擴(kuò)展模塊教程中文版，涵蓋擴(kuò)展模塊安裝、SFM算法、立體視覺、目標(biāo)跟蹤、生物視覺、超分辨率處理等二十多章內(nèi)容。
下載2：Python視覺實(shí)戰(zhàn)項(xiàng)目52講 在「小白學(xué)視覺」公眾號后臺回復(fù)：Python視覺實(shí)戰(zhàn)項(xiàng)目，即可下載包括圖像分割、口罩檢測、車道線檢測、車輛計(jì)數(shù)、添加眼線、車牌識別、字符識別、情緒檢測、文本內(nèi)容提取、面部識別等31個視覺實(shí)戰(zhàn)項(xiàng)目，助力快速學(xué)校計(jì)算機(jī)視覺。
下載3：OpenCV實(shí)戰(zhàn)項(xiàng)目20講 在「小白學(xué)視覺」公眾號后臺回復(fù)：OpenCV實(shí)戰(zhàn)項(xiàng)目20講，即可下載含有20個基于OpenCV實(shí)現(xiàn)20個實(shí)戰(zhàn)項(xiàng)目，實(shí)現(xiàn)OpenCV學(xué)習(xí)進(jìn)階。
交流群
歡迎加入公眾號讀者群一起和同行交流，目前有SLAM、三維視覺、傳感器、自動駕駛、計(jì)算攝影、檢測、分割、識別、醫(yī)學(xué)影像、GAN、算法競賽等微信群（以后會逐漸細(xì)分），請掃描下面微信號加群，備注：”昵稱+學(xué)校/公司+研究方向“，例如：”張三 + 上海交大 + 視覺SLAM“。請按照格式備注，否則不予通過。添加成功后會根據(jù)研究方向邀請進(jìn)入相關(guān)微信群。請勿在群內(nèi)發(fā)送廣告，否則會請出群，謝謝理解~
評論
圖片
表情