BM立体匹配算法得到深度视差图,可填自己双目相机参数,可以得到深度视差图,首先获取双目相机参数,然后图像路径,分割成左右图像经过立体校正、重映射畸变矫正,根据需求修改BM算法参数,最后归一化得到深度视差图。
这是填自己双目相机参数的,官方公开的测试图像的BM立体匹配请看我下一篇。
这是我自己双目相机拍的左目图像,右目不放了,基本一样,经过BM立体匹配算法后得到深度视差图:
下面是代码:
import cv2
import numpy as np
# 基本参数设置
left_camera_matrix = np.array([[830.3436, 5.2274, 577.4470],
[0, 837.6772, 411.9814],
[0, 0, 1]])
right_camera_matrix = np.array([[827.4272, 0.4823, 636.2954],
[0, 836.2356, 405.4773],
[0, 0, 1]])
left_distortion = np.array([[-0.0531, 0.2968, -0.0202, -0.0026, -0.4999]])
right_distortion = np.array([[-0.0408, -0.1898, -0.0217, 0.0029, -0.1983]])
R = np.array([[0.9999, 0.0085, -0.0062],
[-0.0084, 0.9999, 0.0068],
[0.0062, -0.0068, 1.0000]])
T = np.array([[-119.2072], [1.5289], [0.1504]])
size = (1280, 960)
# 畸变矫正和立体校正
R1, R2, P1, P2, Q, validPixROI1, validPixROI2 = cv2.stereoRectify(left_camera_matrix, left_distortion,
right_camera_matrix, right_distortion, size, R, T)
left_map1, left_map2 = cv2.initUndistortRectifyMap(left_camera_matrix, left_distortion, R1, P1, size, cv2.CV_16SC2)
right_map1, right_map2 = cv2.initUndistortRectifyMap(right_camera_matrix, right_distortion, R2, P2, size, cv2.CV_16SC2)
# 加载图片并分割为左右两部分
frame = cv2.imread("D:\chengxu\shuangmu\Stereo-Detection-main\BM(Python)\ccc/frame_0.jpg") # 替换为你的图片路径
frame1 = frame[0:960, 0:1280]
frame2 = frame[0:960, 1280:2560]
# 灰度转换
imgL = cv2.cvtColor(frame1, cv2.COLOR_BGR2GRAY)
imgR = cv2.cvtColor(frame2, cv2.COLOR_BGR2GRAY)
imgL = cv2.remap(imgL, left_map1, left_map2, cv2.INTER_LINEAR)
imgR = cv2.remap(imgR, right_map1, right_map2, cv2.INTER_LINEAR)
# 使用BM算法计算视差图
# BM
numberOfDisparities = ((1280 // 8) + 15) & -16 # 640对应是分辨率的宽
stereo = cv2.StereoBM_create(numDisparities=16, blockSize=9) # 立体匹配
stereo.setPreFilterCap(31)
stereo.setBlockSize(15)
stereo.setMinDisparity(0)
stereo.setNumDisparities(numberOfDisparities)
stereo.setTextureThreshold(10)
stereo.setUniquenessRatio(5)
stereo.setSpeckleWindowSize(100)
stereo.setSpeckleRange(64)
stereo.setDisp12MaxDiff(1)
disparity = stereo.compute(imgL, imgR)
# 视差图归一化
disp = cv2.normalize(disparity, disparity, alpha=0, beta=255, norm_type=cv2.NORM_MINMAX, dtype=cv2.CV_8U)
# 显示视差图
cv2.imshow('Disparity Map', disp)
cv2.waitKey(0)
cv2.destroyAllWindows()
这是填自己双目相机参数的,官方公开的测试图像的BM立体匹配请看我下一篇。