怎么利用已经保存好的stereo_right_mtx.npy,stereo_right_dist.npy,stereo_left_mtx.npy,stereo_left_dist.npy,R_left2right.npy,T_left2right.npy进行双目矫正

你可以使用cv2库中的函数cv2.stereoRectify()，具体方法如下所示：

# 加载矫正参数
right_mtx = np.load('stereo_right_mtx.npy')
right_dist = np.load('stereo_right_dist.npy')
left_mtx = np.load('stereo_left_mtx.npy')
left_dist = np.load('stereo_left_dist.npy')
R_left2right = np.load('R_left2right.npy')
T_left2right = np.load('T_left2right.npy')

# 获取相机的分辨率
h, w = img_shape

# 计算旋转矩阵
R_rec, _, _, _, _, _ = cv2.stereoRectify(
left_mtx, left_dist,
right_mtx, right_dist,
(w, h), R_left2right, T_left2right)

# 计算双目映射表
left_rectify = cv2.initUndistortRectifyMap(
left_mtx, left_dist, None,
R_rec, left_mtx, (w, h), cv2.CV_32FC1)
right_rectify = cv2.initUndistortRectifyMap(
right_mtx, right_dist, None,
R_rec, right_mtx, (w, h), cv2.CV_32FC1)

# 矫正左右图
img_left = cv2.remap(img_left, left_rectify[0], left_rectify[1], cv2.INTER_LINEAR)
img_right = cv2.remap(img_right, right_rectify[0], right_rectify[1], cv2.INTER_LINEAR)

希望可以帮到你！