Revision history [back]

Hi all,

I'm going to try to imitate stereo-imaging by having 1 camera (I am using a fixed cam with 4K res) fixed, and moving the object 3 cm left/right (baseline) and taking two pictures later on feeding them into:

import numpy as np
import cv2
from matplotlib import pyplot as plt

imgL = cv2.imread('left.png',0)
imgR = cv2.imread('right.png',0)

stereo = cv2.StereoBM_create(numDisparities=16, blockSize=15)
disparity = stereo.compute(imgL,imgR)
plt.imshow(disparity,'gray')
plt.show()

However, before I try this, I wanted to consult the vision experts here.

1. Is OpenCV able to construct a depth map from two images taken from a single camera, with a shift of 3 cm or so?
2. If this is possible, what's the accuracy of the depth map? Like, how accurate would it get? Could I perceive depths like 3-4 mm? What kind of custom setup do I need to perceive that little gap differences? How far should the distance to the object be, as well as the baseline? (shift here, basically)

Thanks in advance.

Hi all,

I'm going to try to imitate stereo-imaging by having 1 camera (I am using a fixed cam camera with 4K res) fixed, and moving the object 3 cm left/right (baseline) and taking two pictures later on feeding them into:

import numpy as np
import cv2
from matplotlib import pyplot as plt

imgL = cv2.imread('left.png',0)
imgR = cv2.imread('right.png',0)

stereo = cv2.StereoBM_create(numDisparities=16, blockSize=15)
disparity = stereo.compute(imgL,imgR)
plt.imshow(disparity,'gray')
plt.show()

However, before I try this, I wanted to consult the vision experts here.

1. Is OpenCV able to construct a depth map from two images taken from a single camera, with a shift of 3 cm or so?
2. If this is possible, what's the accuracy of the depth map? Like, how accurate would it get? Could I perceive depths like 3-4 mm? What kind of custom setup do I need to perceive that little gap differences? How far should the distance to the object be, as well as the baseline? (shift here, basically)

Thanks in advance.

Hi all,

I'm going to try to imitate stereo-imaging by having 1 camera (I am using a camera with 4K res) fixed, and moving the object 3 0.8 cm left/right (baseline) and taking two pictures later on feeding them into:

import numpy as np
import cv2
from matplotlib import pyplot as plt

imgL = cv2.imread('left.png',0)
imgR = cv2.imread('right.png',0)

stereo = cv2.StereoBM_create(numDisparities=16, blockSize=15)
disparity = stereo.compute(imgL,imgR)
plt.imshow(disparity,'gray')
plt.show()

However, before I try this, I wanted to consult the vision experts here.my output looks a bit wierd.

Thanks in advance.