November 21, 2019 Development Notes

Changed several bug:

1, the stack by the global variable into a local variable, fixed the starting point of the same problem.

2, the low-pass filter to a high pass filter, the correction of the problem of too much data is pushed onto the stack.

f = np.fft.fft2(new_gray_array)
rows,cols=f.shape
f[0:fftadd]=0
f[rows-fftadd:rows]=0

Question: The new algorithm is influenced by light

 

Use a for loop to write two rectangles, making some good accuracy

leng = len(s.stack)
if leng !=0:
    if(leng > 4):
        for i in range(1, int(leng / 2)+2, int(leng / 2)):
            s_1 = list(s.stack[i])
            s_2 = list(s.stack[i + int(leng / 2)-2])
            x = s_1[0]
            y = s_1[1]
            x_1 = s_2[0]
            y_1 = s_2[1]
            cv2.rectangle(frame, (y, x), (y_1, x_1), (255, 255, 0), 2)  # 绘制矩形
            print(x, y, x_1, y_1)
            print(1)
        cv2.imshow("fps", frame)

    else:
        s_1 = list(s.stack[0])
        s_2 = list(s.stack[leng  - 1])
        x = s_1[0]
        y = s_1[1]
        x_1 = s_2[0]
        y_1 = s_2[1]
        cv2.rectangle(frame, (y, x), (y_1, x_1), (255, 255, 0), 2)  # 绘制矩形
        print(x, y, x_1, y_1)
        cv2.imshow("fps", frame)

Noise is still very large, consider whether or not you want to get some de-noising function to make the image more smooth.

 

Guess you like

Origin www.cnblogs.com/lvfengkun/p/11906661.html