-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathdemo2.py
233 lines (210 loc) · 8.27 KB
/
demo2.py
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
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
import math
import numpy as np
import cv2 as cv
from PIL import ImageFont, ImageDraw, Image
import time
class Detector(object):
def __init__(self):
self.center = [0,0] #表示圆形轨迹的圆心
self.radius = 0 #表示圆形轨迹的半径
self.roix, self.roiy, self.roiw, self.roih = 0, 0, 0, 0
self.color = ""
self.target_ls = []
self.ct = 0 #记录当前帧数
self.start = 0
def add_font(self, img, string, position, font_size):
# 添加文字
font = ImageFont.truetype("font/simsun.ttc", font_size)
img_pil = Image.fromarray(img)
draw = ImageDraw.Draw(img_pil)
draw.text(position, string, font=font, fill=(0, 0, 255))
bk_img = np.array(img_pil)
return bk_img
def calculate_distance(self, box):
# 计算box中前两点之间的距离
distance = math.sqrt((box[0][0] - box[1][0]) ** 2 + math.sqrt((box[0][1] - box[1][1]) ** 2))
return distance
def LeastSquaresCircleFitting(self, points):
# 用于拟合圆形轨迹
sumX, sumY, sumX2, sumY2, sumX3, sumY3, sumXY, sumX1Y2, sumX2Y1 = 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0
N = len(points)
for i in range(N):
x = points[i][0]
y = points[i][1]
x2 = x ** 2
y2 = y ** 2
x3 = x2 * x
y3 = y2 * y
xy = x * y
x1y2 = x * y2
x2y1 = x2 * y
sumX += x
sumY += y
sumX2 += x2
sumY2 += y2
sumX3 += x3
sumY3 += y3
sumXY += xy
sumX1Y2 += x1y2
sumX2Y1 += x2y1
C = N * sumX2 - sumX * sumX
D = N * sumXY - sumX * sumY
E = N * sumX3 + N * sumX1Y2 - (sumX2 + sumY2) * sumX
G = N * sumY2 - sumY * sumY
H = N * sumX2Y1 + N * sumY3 - (sumX2 + sumY2) * sumY
denominator = C * G - D * D
if denominator == 0:
return [0, 0], 0
a = (H * D - E * G) / (denominator)
denominator = D * D - G * C
b = (H * C - E * D) / (denominator)
c = -(a * sumX + b * sumY + sumX2 + sumY2) / N
self.center[0] = a / (-2)
self.center[1] = b / (-2)
self.radius = math.sqrt(a * a + b * b - 4 * c) / 2
def detect_target(self, img):
# 用于检测大符击打目标的函数
target = []
img_roi = img[int(self.roiy):int(self.roiy) + int(self.roih), int(self.roix):int(self.roix) + int(self.roiw)]
hsv = cv.cvtColor(np.array(img_roi), cv.COLOR_BGR2HSV) # 将RGB图转换为灰度图
if (self.color=="red"):
lower_red1 = np.array([0, 43, 46])
lower_red2 = np.array([156, 43, 46])
upper_red1 = np.array([10, 255, 255])
upper_red2 = np.array([180, 255, 255])
mask1 = cv.inRange(hsv, lower_red1, upper_red1) # 根据第一个阈值范围确定mask1
mask2 = cv.inRange(hsv, lower_red2, upper_red2) # 根据第二个阈值范围确定mask2
mask = cv.add(mask1, mask2) # 将两个mask合并
elif (self.color == "blue"):
lower_blue = np.array([100, 43, 46])
upper_blue = np.array([124, 255, 255])
mask = cv.inRange(hsv, lower_blue, upper_blue)
cv.imshow("mask", mask)
kernel = cv.getStructuringElement(cv.MORPH_RECT, (5, 5)) #获得矩形形状的结构元素
closing = cv.morphologyEx(mask, cv.MORPH_CLOSE, kernel) # 对图像进行闭运算
contours, hierarchy = cv.findContours(closing, cv.RETR_TREE, cv.CHAIN_APPROX_SIMPLE) # 对闭运算后的图像寻找轮廓
contour_num_ls = [0 for i in range(len(contours))]
if (hierarchy is None):
return target
hierarchy = hierarchy[0]
for i in range(len(contours)):
if (hierarchy[i][3] != -1): # 表示如果当前轮廓有父轮廓，则将对应的父轮廓值+1
contour_num_ls[hierarchy[i][3]] += 1
box_ls = []
box_distance_ls = []
for i in range(len(contour_num_ls)):
if (contour_num_ls[i] == 1):
choice_index = hierarchy[i][2] # 选取仅有一个子轮廓的父轮廓，并将其子轮廓作为target
rect = cv.minAreaRect(contours[choice_index]) # 最小外接矩形
box = np.int0(cv.boxPoints(rect)) # 矩形的四个角点取整
box_ls.append(box)
box_distance_ls.append(self.calculate_distance(box))
if (len(box_ls) != 0):
#过滤噪声边框
box_distance_ls = np.array(box_distance_ls)
max_index = np.argmax(box_distance_ls)
box = box_ls[max_index]
for j in range(4):
box[j][0] += int(self.roix)
box[j][1] += int(self.roiy)
for j in range(4):
cv.line(img, box[j], box[(j + 1) % 4], color=(0, 255, 0), thickness=2)
target = np.int0((box[0] + box[2]) / 2)
return target
def predict(self, time_gap, rotation):
# 大幅运动预测
a = (0.78+1.045)/2
w = (1.884+2)/2
b = 2.09-a
position_now = self.target_ls[-1] #(x2,y2)
angle_now = math.atan((position_now[1] - self.center[1]) / (position_now[0] - self.center[0]))
delta_angle = ((-a / w) * (-2) * (math.sin(w * 0)) * (math.sin(0.5*w*time_gap))) + b * (
time_gap)
angle_predict = angle_now + delta_angle
if (rotation=="clock"):
if (position_now[0]>self.center[0]):
predict_x = self.center[0] - self.radius * math.cos(angle_predict+math.pi)
predict_y = self.center[1] - self.radius * math.sin(angle_predict+math.pi)
else:
predict_x = self.center[0] - self.radius * math.cos(angle_predict)
predict_y = self.center[1] - self.radius * math.sin(angle_predict)
else:
if (position_now[0] > self.center[0]):
predict_x = self.center[0] + self.radius * math.cos(angle_predict + math.pi)
predict_y = self.center[1] + self.radius * math.sin(angle_predict + math.pi)
else:
predict_x = self.center[0] + self.radius * math.cos(angle_predict)
predict_y = self.center[1] + self.radius * math.sin(angle_predict)
return predict_x, predict_y
def draw_circle(self, img):
if (len(self.target_ls)<5):
return
self.LeastSquaresCircleFitting(self.target_ls)
if (self.center[0] != 0 and self.center[1] != 0 and self.radius != 0):
cv.circle(img, (int(self.center[0]), int(self.center[1])), 5, (255, 0, 0), thickness=cv.FILLED)
cv.circle(img, (int(self.center[0]), int(self.center[1])), int(self.radius), (255, 255, 0), 5)
def get_rotation(self):
# 判断大符的旋转方向
if (len(self.target_ls) < 10):
return ""
temp_target = self.target_ls[len(self.target_ls) - 10:len(self.target_ls)]
angle_ls = [(temp_target[i][0] - self.center[0]) / (temp_target[i][1] - self.center[1]) for i in
range(len(temp_target))]
grad_angle_num_1 = 0
for i in range(len(angle_ls) - 1):
if (angle_ls[i + 1] > angle_ls[i]):
grad_angle_num_1 += 1
if (grad_angle_num_1 > len(angle_ls) - 1 - grad_angle_num_1):
rotation = "anti-clock"
else:
rotation = "clock"
return rotation
def detect(self, file_name, color, time_gap):
# 逐帧读取视频并检测
self.color = color
cap = cv.VideoCapture(file_name)
while True:
success, frame = cap.read() # 逐帧读入
if success:
frame = cv.resize(frame, (800, 600)) # 改变每一帧的大小，使画面适应于电脑屏幕界面
if (self.ct == 0):
# 如果当前是第一帧，则执行框选ROI的操作
roi = cv.selectROI(windowName="fans", img=frame, showCrosshair=True,
fromCenter=False)
self.roix, self.roiy, self.roiw, self.roih = roi
cv.waitKey(-1)
#绘制ROI矩形
cv.rectangle(img=frame, pt1=(self.roix-2, self.roiy-2), pt2=(self.roix-2 + self.roiw+2, self.roiy-2 + self.roih+2), color=(0, 0, 255),
thickness=2)
target = self.detect_target(frame) #返回检测到的击打点位置
if (len(target)!=0):
self.target_ls.append(target)
cv.circle(frame, (int(target[0]), int(target[1])), 5, (0, 0, 255), thickness=cv.FILLED)
print("打", target)
self.draw_circle(frame)
rotation = self.get_rotation()
frame = self.add_font(frame, "rotation:" + rotation, (0, 30), font_size=30)
if (len(self.target_ls)>10):
predict_x, predict_y = self.predict(time_gap, "clock")
cv.circle(frame, (int(predict_x), int(predict_y)), 5, (0, 0, 255), thickness=cv.FILLED)
print("打——预测",[predict_x,predict_y])
# if (len(self.target_ls)>10):
# self.predict(start, end)
# fps = int(1 / (end - start))
fps = cap.get(cv.CAP_PROP_FPS)
frame = self.add_font(frame, "FPS:" + str(int(fps)), (0, 0), font_size=30) # 为当前帧添加帧率
cv.namedWindow('fans', 0)
cv.imshow('fans', frame)
key = cv.waitKey(1) & 0xFF
if key == 27: # print('手动停止')
break
self.ct += 1
else:
print(file_name)
print('播放完成')
break
cap.release()
cv.destroyAllWindows()
if __name__=="__main__":
Detector()
Detector().detect(file_name="red_test1.mp4", color="red", time_gap=0.5)