树莓皮皮卡梅拉-图像比较

Question

我试图检测从覆盆子pi图像中拍摄的两张照片之间的运动。我通过比较原始像素数据来做到这一点。下面是一个简单的脚本，我用它来测试比较。

import picamera
import datetime
from fractions import Fraction
import time
import numpy as np

oldImage = np.empty((48, 64, 3))
newImage = np.empty((48, 64, 3))
color_offset = 25

with picamera.PiCamera() as camera:
    camera.resolution = (64,48)
    camera.start_preview()
    time.sleep(2)

    camera.capture(oldImage, "rgb")
    time.sleep(5)
    camera.capture(newImage, "rgb")

    x = 0
    y = 0
    diff = 0

    while(x < np.size(newImage, 0)):
        while(y < np.size(newImage, 1)):
            val1 = newImage[x, y, 0] + newImage[x, y, 1] + newImage[x, y, 2]
            val2 = oldImage[x, y, 0] + oldImage[x, y, 1] + oldImage[x, y, 2]
            print(val1)
            print(val2)
            pd = abs(val2 - val1)
            print(pd)

            if(pd > color_offset):
                diff += 1
            y +=1
        x += 1
        y=0

    print(str(diff))

在脚本中，我启动相机，拍摄第一张照片，等待几秒钟，拍摄第二张照片并计算差值。但是每次我运行这个脚本时，“val1”变量总是0.0。任何帮助都将不胜感激。

Answer 1

所以用2x2的“图像”，你的颜值加成

import numpy as np

newImage = np.array([[[0,0,0], [1,0,0]], [[0,1,0], [0,1,1]]])

x = 0
y = 0
while(x < np.size(newImage, 0)):
    while(y < np.size(newImage, 1)):
        val1 = newImage[x, y, 0] + newImage[x, y, 1] + newImage[x, y, 2]
        print(val1)
        y +=1
    x += 1
    y = 0

生产

0
1
1
2

正如我所料。

你期望得到什么结果？您的数据是否正确（提供样本数据）？