1#!/usr/bin/python 2""" 3 Tracking of rotating point. 4 Rotation speed is constant. 5 Both state and measurements vectors are 1D (a point angle), 6 Measurement is the real point angle + gaussian noise. 7 The real and the estimated points are connected with yellow line segment, 8 the real and the measured points are connected with red line segment. 9 (if Kalman filter works correctly, 10 the yellow segment should be shorter than the red one). 11 Pressing any key (except ESC) will reset the tracking with a different speed. 12 Pressing ESC will stop the program. 13""" 14import cv2 15from math import cos, sin 16import numpy as np 17 18if __name__ == "__main__": 19 20 img_height = 500 21 img_width = 500 22 kalman = cv2.KalmanFilter(2, 1, 0) 23 24 code = -1L 25 26 cv2.namedWindow("Kalman") 27 28 while True: 29 state = 0.1 * np.random.randn(2, 1) 30 31 kalman.transitionMatrix = np.array([[1., 1.], [0., 1.]]) 32 kalman.measurementMatrix = 1. * np.ones((1, 2)) 33 kalman.processNoiseCov = 1e-5 * np.eye(2) 34 kalman.measurementNoiseCov = 1e-1 * np.ones((1, 1)) 35 kalman.errorCovPost = 1. * np.ones((2, 2)) 36 kalman.statePost = 0.1 * np.random.randn(2, 1) 37 38 while True: 39 def calc_point(angle): 40 return (np.around(img_width/2 + img_width/3*cos(angle), 0).astype(int), 41 np.around(img_height/2 - img_width/3*sin(angle), 1).astype(int)) 42 43 state_angle = state[0, 0] 44 state_pt = calc_point(state_angle) 45 46 prediction = kalman.predict() 47 predict_angle = prediction[0, 0] 48 predict_pt = calc_point(predict_angle) 49 50 measurement = kalman.measurementNoiseCov * np.random.randn(1, 1) 51 52 # generate measurement 53 measurement = np.dot(kalman.measurementMatrix, state) + measurement 54 55 measurement_angle = measurement[0, 0] 56 measurement_pt = calc_point(measurement_angle) 57 58 # plot points 59 def draw_cross(center, color, d): 60 cv2.line(img, 61 (center[0] - d, center[1] - d), (center[0] + d, center[1] + d), 62 color, 1, cv2.LINE_AA, 0) 63 cv2.line(img, 64 (center[0] + d, center[1] - d), (center[0] - d, center[1] + d), 65 color, 1, cv2.LINE_AA, 0) 66 67 img = np.zeros((img_height, img_width, 3), np.uint8) 68 draw_cross(np.int32(state_pt), (255, 255, 255), 3) 69 draw_cross(np.int32(measurement_pt), (0, 0, 255), 3) 70 draw_cross(np.int32(predict_pt), (0, 255, 0), 3) 71 72 cv2.line(img, state_pt, measurement_pt, (0, 0, 255), 3, cv2.LINE_AA, 0) 73 cv2.line(img, state_pt, predict_pt, (0, 255, 255), 3, cv2.LINE_AA, 0) 74 75 kalman.correct(measurement) 76 77 process_noise = kalman.processNoiseCov * np.random.randn(2, 1) 78 state = np.dot(kalman.transitionMatrix, state) + process_noise 79 80 cv2.imshow("Kalman", img) 81 82 code = cv2.waitKey(100) % 0x100 83 if code != -1: 84 break 85 86 if code in [27, ord('q'), ord('Q')]: 87 break 88 89 cv2.destroyWindow("Kalman") 90