我有一个图像,我想基于图像的value-channel
像素的xy
的值在其上放置遮罩。
def rgb_mask(img):
r, g, b = img[:,:,2], img[:,:,1], img[:,:,0]
intensity = cv2.cvtColor(frame, cv2.COLOR_BGR2HSV)[:,:,2]
mask_if_intensity_below_50 = (np.logical_and((0.85 * g > 0.95 * b), (0.95 * g > 0.85 * r)) * 255).astype(np.uint8)
mask_if_intensity_between_50_and_200 = (np.logical_and((0.85 * g > 0.95 * b), (0.95 * g > 0.85 * r), (g - r + b > 30)) * 255).astype(np.uint8)
mask_if_intensity_above_200 = (np.logical_and((0.85 * g > 0.95 * b), (0.95 * g > 0.85 * r), (g - b < 150)) * 255).astype(np.uint8)
masked = cv2.bitwise_and(img, img, mask=?) # I typed ? because I am note sure how to include this in the code
return masked
img.shape
返回以下内容:
(720, 1280, 3)
如何为每个像素分配正确的遮罩?理想情况下,我不想使用for x, for y
循环。在此先感谢
bitwise_and
之前将遮罩复制为3D矩阵。 您可以使用numpy dstak函数复制蒙版。例如:np.dstack((msk, msk, msk))
的形状为msk(720,1280),返回形状为(720,1280,3)
在以下示例中,我假设intensity
矩阵在[0,255]范围内(而不是[0,1]):
mask_if_intensity_between_100_and_200 = (np.logical_and(intensity > 100, intensity < 200) * 255).astype(np.uint8)
masked = cv2.bitwise_and(img, np.dstack((mask_if_intensity_between_100_and_200, mask_if_intensity_between_100_and_200, mask_if_intensity_between_100_and_200)))
当遮罩形状和图像形状相同时,可以按位和。