SIFT test script

sift_example.py

import numpy as np
import cv2
from matplotlib import pyplot as plt

def show_img(img1, img2, p1, p2, status):
    h1, w1 = img1.shape[:2]
    h2, w2 = img2.shape[:2]
    vis = np.zeros((max(h1, h2), w1+w2), np.uint8)
    vis[:h1, :w1] = img1
    vis[:h2, w1:w1+w2] = img2
    vis = cv2.cvtColor(vis, cv2.COLOR_GRAY2BGR)
    p2 += (w1, 0)
    # plot the matches
    color = (0, 255, 0)
    for (x1, y1), (x2, y2), inlier in zip(p1, p2, status):
        if inlier:
            cv2.circle(vis, (x1, y1), 2, color, -1)
            cv2.circle(vis, (x2, y2), 2, color, -1)
            cv2.line(vis, (x1, y1), (x2, y2), color)
    plt.imshow(vis)
    plt.axis('off')
    plt.show()

def main():
    """
    execute this function when the script is ran.
    """
    sign = cv2.imread('signs_3.png')
    img = cv2.imread('a3.jpg')

    sign = cv2.cvtColor(sign, cv2.COLOR_BGR2GRAY)
    img = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)

    detector = cv2.SURF()
    sign_kp, sign_desc = detector.detectAndCompute(sign, None)
    img_kp, img_desc = detector.detectAndCompute(img, None)

    matcher = cv2.BFMatcher()
    matches = matcher.knnMatch(sign_desc, trainDescriptors=img_desc, k=2)

    #filter matches
    matches = [(match[0].queryIdx, match[1].trainIdx) for match in matches \
        if match[0].distance < match[1].distance * 0.75]

    sign_pts = np.float32([sign_kp[match[0]].pt for match in matches])
    img_pts = np.float32([img_kp[match[1]].pt for match in matches])

    _, status = cv2.findHomography(sign_pts, img_pts, cv2.RANSAC, 5.0)
    show_img(sign, img, sign_pts, img_pts, status)
    #for all desc  matches get the keypoints

if __name__ == "__main__":
    main()

A note to self do not use SIFT for shape detection, try template matching.