python图像特征提取

这里使用的是python 3.6 、opencv_python-3.4.0+contrib，特征提取的代码如下：

import cv2

img = cv2.imread("feature.jpg")
img = cv2.resize(img,(136 * 3,76 * 3))
cv2.imshow("original",img)

gray = cv2.cvtColor(img,cv2.COLOR_RGB2GRAY)

#使用SIFT
sift = cv2.xfeatures2d.SIFT_create()                    
keypoints, descriptor = sift.detectAndCompute(gray,None)

cv2.drawKeypoints(image = img,
                  outImage = img,
                  keypoints = keypoints,
                  flags = cv2.DRAW_MATCHES_FLAGS_DRAW_RICH_KEYPOINTS,
                  color = (51,163,236))
cv2.imshow("SIFT",img)

#使用SURF
img = cv2.imread("feature.jpg")
img = cv2.resize(img,(136 * 3,76 * 3))

surf = cv2.xfeatures2d.SURF_create()
keypoints, descriptor = surf.detectAndCompute(gray,None)

cv2.drawKeypoints(image = img,
                  outImage = img,
                  keypoints = keypoints,
                  flags = cv2.DRAW_MATCHES_FLAGS_DRAW_RICH_KEYPOINTS,
                  color = (51,163,236))
cv2.imshow("SURF",img)

img = cv2.imread("feature.jpg")
img = cv2.resize(img,(136 * 3,76 * 3))

cv2.waitKey(0)
cv2.destroyAllWindows()

结提取果：

keypoints, descriptor = sift.detectAndCompute(gray,None)，这里的descriptor是一个向量或者应该可以转换成向量