python cv2 基於surf特徵點匹配的圖像拼接

基於surf特徵點匹配的圖像拼接:

1.讀取要拼接的左右圖像;

2.提取左右圖像的surf特徵點;

3.利用knn對左右圖像的特徵點進行匹配;

4.利用匹配的特徵對，求出單應矩陣;

5.利用單應矩陣對右圖像進行透視變化;

6.對左右圖像進行拼接。

具體如下代碼：

import cv2
import numpy as np
def warp_corner(H,src):
    '''
    :param H: 單應矩陣
    :param src: 透視變化的圖像
    :return: 透視變化後的四個角，左上角開始，逆時鐘
    '''

    warp_points=[]
    # 圖像左上角，左下角
    src_left_up=np.array([0,0,1])
    src_left_down=np.array([0,src.shape[0],1])

    # 圖像右上角，右下角
    src_right_up=np.array([src.shape[1],0,1])
    src_right_down=np.array([src.shape[1],src.shape[0],1])

    #透視變化後的左上角，左下角
    warp_left_up=H.dot(src_left_up)
    left_up=warp_left_up[0:2]/warp_left_up[2]
    warp_points.append(left_up)
    warp_left_down=H.dot(src_left_down)
    left_down=warp_left_down[0:2]/warp_left_down[2]
    warp_points.append(left_down)

    # 透視變化後的右上角，右下角
    warp_right_up=H.dot(src_right_up)
    right_up=warp_right_up[0:2]/warp_right_up[2]
    warp_points.append(right_up)
    warp_right_down=H.dot(src_right_down)
    right_down=warp_right_down[0:2]/warp_right_down[2]
    warp_points.append(right_down)
    return warp_points
def optim_mask(mask,warp_point):


    min_left_x = min(warp_point[0][0], warp_point[1][0])
    left_margin = mask.shape[1] - min_left_x
    points_zeros = np.where(mask == 0)
    x_indexs = points_zeros[1]
    alpha = (left_margin - (x_indexs - min_left_x)) / left_margin
    mask[points_zeros] = alpha
    return mask
def Seam_Left_Right(left,imagewarp,H,warp_point,with_optim_mask=False):
    '''
    :param left: 拼接的左圖像
    :param imagewarp: 透視變化後的右圖像
    :param H: 單應矩陣
    :param warp_point: 透視變化後的四個頂點
    :param with_optim_mask: 是否需要對拼接後的圖像進行優化
    :return:
    '''
    w = left.shape[1]
    mask = imagewarp[:, 0:w]
    mask = cv2.cvtColor(mask, cv2.COLOR_BGR2GRAY)
    mask[mask != 0] = 1
    mask[mask == 0] = 0
    mask = 1 - mask
    mask = np.float32(mask)

    if with_optim_mask==True:
        mask=optim_mask(mask,warp_point)
    mask_rgb=np.stack([mask,mask,mask],axis=2)
    tt=np.uint8((1-mask_rgb)*255)
    left=left*mask_rgb+imagewarp[:,0:w]*(1-mask_rgb)
    imagewarp[:,0:w]=left
    return np.uint8(imagewarp)
def main():
    left=cv2.imread('/home/simple/convid_withornot_mask/recogizition_withornot_mask/ttt/b_t.jpg')
    #left=cv2.resize(left,dsize=(512,512))
    left_gray=cv2.cvtColor(left,cv2.COLOR_BGR2GRAY)
    right=cv2.imread('/home/simple/convid_withornot_mask/recogizition_withornot_mask/ttt/b.jpg')
    #right=cv2.resize(right,dsize=(512,512))
    right_gray=cv2.cvtColor(right,cv2.COLOR_BGR2GRAY)

    #提取左右圖像的surf特徵點
    detector=cv2.xfeatures2d_SURF.create(hessianThreshold=400)
    left_kps,left_dess=detector.detectAndCompute(left_gray,None)
    right_kps,right_dess=detector.detectAndCompute(right_gray,None)

    #利用knn對左右圖像的特徵點進行匹配
    matcher=cv2.FlannBasedMatcher_create()
    knn_matchers=matcher.knnMatch(left_dess,right_dess,2)
    good_keypoints=[]


    #挑出好的匹配點
    for m,n in knn_matchers:
        if m.distance<0.5*n.distance:
            good_keypoints.append(m)
    left_points=np.zeros(shape=(len(good_keypoints),2),dtype=np.float32)
    right_points=np.zeros(shape=(len(good_keypoints),2),dtype=np.float32)
    outimg=np.zeros(shape=(right.shape[0],right.shape[0]+left.shape[0],3),dtype=np.uint8)
    cv2.drawMatches(left,left_kps,right,right_kps,good_keypoints,outimg)
    # cv2.imshow('hks',outimg)
    # cv2.waitKey(0)
    for i in range(len(good_keypoints)):
        left_points[i][0]=left_kps[good_keypoints[i].queryIdx].pt[0]
        left_points[i][1]=left_kps[good_keypoints[i].queryIdx].pt[1]
        right_points[i][0]=right_kps[good_keypoints[i].trainIdx].pt[0]
        right_points[i][1]=right_kps[good_keypoints[i].trainIdx].pt[1]

    #求取單應矩陣
    H,_=cv2.findHomography(right_points,left_points)

    #求出右圖像的透視變化頂點
    warp_point = warp_corner(H, right)
    #求出右圖像的透視變化圖像
    imagewarp=cv2.warpPerspective(right,H,(left.shape[1]+right.shape[1],left.shape[0]))

    #對左右圖像進行拼接，返回最後的拼接圖像
    image_seam_optim=Seam_Left_Right(left,imagewarp,H,warp_point,with_optim_mask=True)
    cv2.namedWindow('image_seam_optim',cv2.WINDOW_NORMAL)
    cv2.imshow('image_seam_optim',image_seam_optim)
    cv2.waitKey(0)
if __name__=='__main__':
    main()

其中，圖像拼接時，有對結果進行優化，具體參考代碼。

效果如下：

                                                                                                                               左圖

 

                                                                                                                                    右圖

                                                                                                                  拼接的圖像

                                                                                                                            優化後的圖像拼接