错误的视差图

Question

我正在做一个障碍物探测器项目，并且正在使用立体声相机。我已经完成了两个相机的校准。但是对于计算视差，我感到视差图不理想。视差图的图像在这里：

我认为在将参数传递给cv.StereoSGBM_create（）函数时犯了一些错误。谁能解释我如何获得正确的视差图？我在python中的代码在下面

from __future__ import print_function

import numpy as np
import cv2 as cv
import glob

ply_header = '''ply
format ascii 1.0
element vertex %(vert_num)d
property float x
property float y
property float z
property uchar red
property uchar green
property uchar blue
end_header
'''

def write_ply(fn, verts, colors):
    verts = verts.reshape(-1, 3)
    colors = colors.reshape(-1, 3)
    verts = np.hstack([verts, colors])
    with open(fn, 'wb') as f:
        f.write((ply_header % dict(vert_num=len(verts))).encode('utf-8'))
        np.savetxt(f, verts, fmt='%f %f %f %d %d %d ')


if __name__ == '__main__':
    print('loading images...')
    imgL = cv.pyrDown( cv.imread('D:/Images/left_img.jpg') )  # downscale images for faster processing
    imgR = cv.pyrDown( cv.imread('D:/Images/right_img.jpg') )
    hl, wl = imgL.shape[:2]
    hr, wr = imgR.shape[:2]
    #reading camera parameters after calibration
    Rc = np.load('D:/Images/Rcalib.npz')
    Lc = np.load('D:/Images/Lcalib.npz')
    # undistorting images
    newCameraMtxR, roiR = cv.getOptimalNewCameraMatrix(Rc['mtx'], Rc['dist'], (wr, hr), 1, (wr, hr))
    udImgR = cv.undistort(imgR, Rc['mtx'], Rc['dist'], None, newCameraMtxR)  #undistorted Right Image
    newCameraMtxL, roiL = cv.getOptimalNewCameraMatrix(Lc['mtx'], Lc['dist'], (wl, hl), 1, (wl, hl))
    udImgL = cv.undistort(imgL, Lc['mtx'], Lc['dist'], None, newCameraMtxL)


    window_size = 3
    min_disp = 16     
    num_disp = 112-min_disp    
    stereo = cv.StereoSGBM_create(minDisparity = min_disp,
        numDisparities = num_disp,
        blockSize = 16,
        P1 = 8*3*window_size**2,
        P2 = 32*3*window_size**2,
        disp12MaxDiff = 1,
        uniquenessRatio = 10,
        speckleWindowSize = 100,
        speckleRange = 32
    )

    print('computing disparity...')
    disp = stereo.compute(udImgL, udImgR).astype(np.float32) / 16.0

    print('generating 3d point cloud...',disp[0][0])
    h, w = udImgL.shape[:2]

    Q = np.float32([[1, 0, 0, -0.5*w],
                    [0,-1, 0,  0.5*h], # turn points 180 deg around x-axis,
                    [0, 0, 0,     -948], # 948 is focal length of cameras from camera matrix,after calibration
                    [0, 0, 1,      0]])
    points = cv.reprojectImageTo3D(disp, Q)
    colors = cv.cvtColor(udImgL, cv.COLOR_BGR2RGB)
    mask = disp > disp.min()     #boolean mask of 2D after condition,where disparity is greater then minimum disparity
    out_points = points[mask]     # contain only those points where this above condition match
    out_colors = colors[mask]
    out_fn = 'out.ply'
    write_ply('out.ply', out_points, out_colors)

    print('%s saved' % 'out.ply')



    cv.imshow('left', udImgL)
    cv.imshow('disparity', (disp-disp.min())/(disp.max()-disp.min()))
    b=0.0635 #baseline
    D=b*948/disp #depth for each point

    print('Max dasparity: ',disp.max())
    mat=np.matrix(D)
    print('distance: ',mat.min())
    with open('outfile.txt','wb') as f:
        for line in mat:
            np.savetxt(f, line, fmt='%.2f')


    cv.waitKey()
    cv.destroyAllWindows()