我正在尝试使用python包装器为OpenCV(cv2)构建一个用于2D对象的veeery简单跟踪器。
我只注意到3个功能:
我的想法是创建一个代码来检查kalman是否像这样工作:
kf = cv2.KalmanFilter(...)
# set initial position
cv2.predict()
corrected_position = cv2.correct([measurement_x, measurement_y])
我发现了一些使用cv包装器而不是cv2 ...
的例子提前致谢!
答案 0 :(得分:28)
如果您正在使用opencv2.4,那么它就是坏消息:KalmanFilter无法使用,因为您无法设置过渡(或任何其他)矩阵。
对于opencv3.0它可以正常工作,如下所示:
import cv2, numpy as np
meas=[]
pred=[]
frame = np.zeros((400,400,3), np.uint8) # drawing canvas
mp = np.array((2,1), np.float32) # measurement
tp = np.zeros((2,1), np.float32) # tracked / prediction
def onmouse(k,x,y,s,p):
global mp,meas
mp = np.array([[np.float32(x)],[np.float32(y)]])
meas.append((x,y))
def paint():
global frame,meas,pred
for i in range(len(meas)-1): cv2.line(frame,meas[i],meas[i+1],(0,100,0))
for i in range(len(pred)-1): cv2.line(frame,pred[i],pred[i+1],(0,0,200))
def reset():
global meas,pred,frame
meas=[]
pred=[]
frame = np.zeros((400,400,3), np.uint8)
cv2.namedWindow("kalman")
cv2.setMouseCallback("kalman",onmouse);
kalman = cv2.KalmanFilter(4,2)
kalman.measurementMatrix = np.array([[1,0,0,0],[0,1,0,0]],np.float32)
kalman.transitionMatrix = np.array([[1,0,1,0],[0,1,0,1],[0,0,1,0],[0,0,0,1]],np.float32)
kalman.processNoiseCov = np.array([[1,0,0,0],[0,1,0,0],[0,0,1,0],[0,0,0,1]],np.float32) * 0.03
#kalman.measurementNoiseCov = np.array([[1,0],[0,1]],np.float32) * 0.00003
while True:
kalman.correct(mp)
tp = kalman.predict()
pred.append((int(tp[0]),int(tp[1])))
paint()
cv2.imshow("kalman",frame)
k = cv2.waitKey(30) &0xFF
if k == 27: break
if k == 32: reset()