TwoColorTracking/TrackingUtility.py at master · chapter544/TwoColorTracking · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
import numpy as np
from ReadCyanSystemsVideo import ScaleUint16To255
import cv2


def BuildBackgroundModels(lw, mw, numHistoryFrames=100, \
                          varThreshold1=5.0, varThreshold2=3.0, \
                          detectShadows=False, \
                          gamma_lw=.9, gamma_mw=0.05):
    lw_max = np.amax(lw)
    mw_max = np.amax(mw)
    lw_min = np.amin(lw)
    mw_min = np.amin(mw)
    lw_range = np.amax(lw) - np.amin(lw)
    mw_range = np.amax(mw) - np.amin(mw)

    # create background subtraction model
    fgbg1 = cv2.createBackgroundSubtractorKNN(numHistoryFrames, varThreshold1, detectShadows)
    fgbg2 = cv2.createBackgroundSubtractorKNN(numHistoryFrames, varThreshold2, detectShadows)
    frameCounter = 0
    while True:
        frameCounter = frameCounter + 1
        if frameCounter > numHistoryFrames:
            break

        lw_frame = ScaleUint16To255(lw[frameCounter], lw_max, lw_min, gamma_lw)
        mw_frame = ScaleUint16To255(mw[frameCounter], mw_max, mw_min, gamma_mw)
        lw_fg = fgbg1.apply(lw_frame)
        mw_fg = fgbg2.apply(mw_frame)

    return fgbg1, fgbg2


def GetTargetObservation(target_cts, pred_x):
    if not target_cts:
        return pred_x

    obj_features = np.zeros( (4, len(target_cts)) )
    for idx, c in enumerate(target_cts):
        # bounding box with minimum area, so it consider rotation.
        # minAreaRect returns a Box2D structure
		# (topleftcorner(x,y), (width, height), angle)
        rect = cv2.minAreaRect(c)
        #print rect
        top_corner = rect[0]
        obj_dim = rect[1]
        #print top_corner
        #print obj_dim
        obj_features[:,idx] = np.array([top_corner[0], \
                                        top_corner[1] , \
                                        obj_dim[0], obj_dim[1]])
        #print top_corner, obj_dim

    #print("Feature obj")
    #print obj_features
    #dist_xy_list = []
    #dist_wh_list = []
    return obj_features


def GetTargetCorrespondent(obj_features, pred_x, xy_dist_max=50.,
        wh_dist_max=20.):
    DISTANCE_TYPES = 2 # xy and wh

    #print obj_features
    dist_lw = np.zeros((DISTANCE_TYPES,obj_features.shape[1]))
    for idx in range(obj_features.shape[1]):
        temp_features = obj_features[:,idx]
        dist_lw[0,idx]= abs(pred_x[0] - temp_features[0]) + \
                  abs(pred_x[2] - temp_features[1])
        dist_lw[1,idx] = abs(pred_x[4] - temp_features[2]) + \
                  abs(pred_x[5] - temp_features[3])

    #print dist_lw
    xy_min_idx = np.argmin(dist_lw[0,:])
    wh_min_idx = np.argmin(dist_lw[1,:])
    #print('xy index %d' % xy_min_idx)
    #print('wh ndex %d' % wh_min_idx)

    is_valid = True
    if dist_lw[0,xy_min_idx] > xy_dist_max or \
            dist_lw[1,wh_min_idx] > wh_dist_max:
        is_valid = False

    xy_err = dist_lw[0, xy_min_idx]
    return obj_features[:,xy_min_idx], is_valid, xy_err


def GetTargetCorrespondentTwoColor(lw_features, mw_features, pred_x):
    lw_feature, lw_valid, lw_err = GetTargetCorrespondent(lw_features, pred_x,
            xy_dist_max=30.,
            wh_dist_max=20.)
    mw_feature, mw_valid, mw_err = GetTargetCorrespondent(mw_features, pred_x,
            xy_dist_max=30.,
            wh_dist_max=20.)

    z_valid = not (lw_valid and mw_valid)
    selected_feature = []

    if z_valid:
        selected_feature = lw_feature
        if lw_err > mw_err:
            selected_feature = mw_feature

    #if lw_valid and not mw_valid:
    #    z_valid = True
    #    selected_feature = lw_feature

    #if not mw_valid and mw_valid:
    #    z_valid = True
    #    selected_feature = mw_feature
    return selected_feature, z_valid


def BackgroundSubtraction(frame, fgbg, kernel, validRegionArea):
    fgmask = fgbg.apply(frame)

    _, thresh = cv2.threshold(fgmask, 127, 255, 0)
    fgmask = cv2.morphologyEx(thresh, cv2.MORPH_CLOSE, kernel)
    fgmask = cv2.morphologyEx(fgmask, cv2.MORPH_CLOSE, kernel)
    fgmask = cv2.morphologyEx(fgmask, cv2.MORPH_CLOSE, kernel)
    fgmask = cv2.morphologyEx(fgmask, cv2.MORPH_CLOSE, kernel)

    #cv2.imwrite("test.jpg", fgmask)

    ret, cts, hier = cv2.findContours(fgmask, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
    cts_sorted = sorted(cts, key = cv2.contourArea, reverse=True)

    good_cts = []
    for c in cts_sorted:
        area = cv2.contourArea(c)
        if area > validRegionArea:
            good_cts.append(c)

    return good_cts


def GetContoursFromBackgroundMask(fgmask, validRegionArea):
    ret, cts, hier = cv2.findContours(fgmask, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
    cts_sorted = sorted(cts, key = cv2.contourArea, reverse=True)

    good_cts = []
    for c in cts_sorted:
        area = cv2.contourArea(c)
        if area > validRegionArea:
            good_cts.append(c)

    return good_cts


def GetBackgroundSubtraction(frame, fgbg, kernel):
    fgmask = fgbg.apply(frame)

    _, thresh = cv2.threshold(fgmask, 127, 255, 0)
    fgmask = cv2.morphologyEx(thresh, cv2.MORPH_CLOSE, kernel)
    fgmask = cv2.morphologyEx(fgmask, cv2.MORPH_CLOSE, kernel)
    fgmask = cv2.morphologyEx(fgmask, cv2.MORPH_CLOSE, kernel)
    fgmask = cv2.morphologyEx(fgmask, cv2.MORPH_CLOSE, kernel)
    fgmask = cv2.morphologyEx(fgmask, cv2.MORPH_CLOSE, kernel)
    fgmask = cv2.morphologyEx(fgmask, cv2.MORPH_CLOSE, kernel)

    return fgmask