3D Stereo Depth Vision with 2 Cameras - OpenCV Python Tutorial

Опубликовано: 16 Май 2026
на канале: Consulting Joe

9,912

149

Check out http://ConsultingJoe.com for more.

::::::::::::::::PYTHON CODE::::::::::::::::

import cv2
import numpy as np
import cvzone
from cvzone.SelfiSegmentationModule import SelfiSegmentation

width = 640
height = 640

Define the cameras to use (camera_1, camera_2)
camera_1 = cv2.VideoCapture(0, cv2.CAP_DSHOW)
camera_1.set(cv2.CAP_PROP_FRAME_WIDTH, width)
camera_1.set(cv2.CAP_PROP_FRAME_HEIGHT, height)
camera_1.set(cv2.CAP_PROP_FPS, 15)

camera_2 = cv2.VideoCapture(1, cv2.CAP_DSHOW)
camera_2.set(cv2.CAP_PROP_FRAME_WIDTH, width)
camera_2.set(cv2.CAP_PROP_FRAME_HEIGHT, height)
camera_2.set(cv2.CAP_PROP_FPS, 15)

img_left = cv2.imread('1.jpg')
img_right = cv2.imread('r1.jpg')

segmentor = SelfiSegmentation(0)

Create a stereo block matching object
stereo = cv2.StereoBM_create(numDisparities=16, blockSize=11)
stereo = cv2.StereoSGBM_create(
minDisparity=12,
numDisparities=64,
blockSize=21)

while True:
camera_1.grab()
camera_2.grab()

Read the frames from each camera
ret1, frame1 = camera_1.read()
ret2, frame2 = camera_2.read()

frame1 = segmentor.removeBG(frame1, (0,0,0))
frame2 = segmentor.removeBG(frame2, (0,0,0))

Convert the frames to grayscale
gray1 = cv2.cvtColor(frame1, cv2.COLOR_BGR2GRAY)
gray2 = cv2.cvtColor(frame2, cv2.COLOR_BGR2GRAY)

kernel = np.ones((3,3), np.float32)/9
gray1 = cv2.filter2D(gray1, -1, kernel)
gray2 = cv2.filter2D(gray2, -1, kernel)

Compute the depth map using the stereo block matching object
depth_map = stereo.compute(gray1, gray2)
Normalize the depth map to be between 0 and 255
norm_depth = cv2.normalize(depth_map, None, 10, 245, cv2.NORM_MINMAX)

Convert the normalized depth map to color
color_depth = cv2.applyColorMap(norm_depth.astype(np.uint8), cv2.COLORMAP_HSV)
color_depth = cv2.cvtColor(color_depth, cv2.COLOR_BGR2GRAY)

Stack the original frames vertically, along with the color depth map
result = np.hstack((frame1, frame2, norm_depth))
color_depth = color_depth[color_depth.shape[0]//2:color_depth.shape[0]]
result = result[:, :-300]
color_depth = cv2.flip(color_depth, 1)
color_depth = color_depth[:height, :width-100]

Display the stacked frames and depth map
cv2.imshow("3D Video", color_depth)
cv2.imshow("Stereo Video and Depth Map", cv2.flip(result, 1))
cv2.waitKey(10000)

Break the loop if the 'q' key is pressed
if cv2.waitKey(1) & 0xFF == ord('q'):
break

Release the cameras
camera_1.release()
camera_2.release()