panorama_test.py

import os
import cv2 as cv
import numpy as np
import matplotlib.pyplot as plt

from stitch import transform
from stitch import utils, Stitch, simpleStitch

pts = []
dst = []

def select_pts(event,x,y,flags,param):
    if event == cv.EVENT_LBUTTONDOWN:
        global pts
        cv.circle(img_stitch_,(x,y),15,(200,155,0),-1)
        pts.append([x,y])

def select_dst(event,x,y,flags,param):
    if event == cv.EVENT_LBUTTONDOWN:
        global dst
        cv.circle(img_stitch_,(x,y),15,(0,200,155),-1)
        dst.append([x,y])
        
# Define the main function
if __name__ == '__main__':
    # Define the draw parameters for matching visualization
    draw_params = dict(matchColor=(0, 255, 0),
                       singlePointColor=(0, 0, 255),
                       flags=cv.DrawMatchesFlags_DEFAULT)
    '''
    Stitch Multiple Images for NISwGSP-denny dataset
    
    Generate estimated homography matrix using matches between corresponding features in separate ROIs.
    Directly apply homography matrix to stitch all frames
    --------------------
    
    :param  img1~img15: original undistorted images 
    :type  img1_~img15_: nparray
    
    :param  homo_mat: homography matrix generated from feature estimation, using for image transformation
    :type  homo_mat: nparray (3*3)
    
    :return  img_stitch: stitched image for 15 images
    '''
    # Load the folder path
    path = "dataset/example_image/APAP-conssite"
    files= os.listdir(path)
    keys = []
    for file in files:
        if not os.path.isdir(file) and not '.mat' in file:
            keys.append(file)
    keys = sorted(keys)
    
    # Load all the images from right to the left
    imgs = {}
    for key in keys:
        img = cv.imread(path+'/'+key)
        imgs[key] = cv.resize(img,(960,720))

    num_images = len(keys)

    for i in range(num_images):
        img2 = imgs[keys[i+1]]
        if i == 0:
            img1 = imgs[keys[i]]
        else:
            cv.namedWindow("Area Cutting", cv.WINDOW_NORMAL)
            cv.resizeWindow("Area Cutting", 1960, 1080)
            ROI = cv.selectROIs("Area Cutting", img_stitch)
            img1 = img_stitch[ROI[0,1]:ROI[0,1]+ROI[0,3],ROI[0,0]:ROI[0,0]+ROI[0,2],:]
            cv.imwrite("img_stitch.png",img1)
        
        # Manually define the ROI to locate the area for corresponding images
        cv.namedWindow("Area Selection", cv.WINDOW_NORMAL)
        cv.resizeWindow("Area Selection", 1960, 1080)
        ROIs1 = cv.selectROIs("Area Selection", img1)
        ROIs2 = cv.selectROIs("Area Selection", img2)

        for i in range(len(ROIs1)):
            ROIs1[i, 2] = ROIs1[i, 0] + ROIs1[i, 2]
            ROIs1[i, 3] = ROIs1[i, 1] + ROIs1[i, 3]
            ROIs2[i, 2] = ROIs2[i, 0] + ROIs2[i, 2]
            ROIs2[i, 3] = ROIs2[i, 1] + ROIs2[i, 3]
            
        # Initialize the object
        stitch = Stitch(img1, img2)
        stitch.featureExtract(ROIs1, ROIs2)


        '''Match the Corresponding Features'''
        # Define the matches based on two images
        matches_list = stitch.clusterMatch('sift', knn=True)

        # Show the number of matches
        matchNum = 0
        for i in range(len(matches_list)):
            matchNum += len(matches_list[i])
            print("-- Number of original matches in each area", len(matches_list[i]))
        print("Number of original total matches: ", matchNum)

        # Combine the features in one lists from each cluster
        stitch.featureIntegrate(matches_list)
        
        '''
        Find the parameters for homography matrix
        '''
        # Calculate the homography matrix for image transformation
        homo_mat, matches_inliers = stitch.homoEstimate()
        img_inliers = cv.drawMatches(img1,stitch.Img1.kps,img2,stitch.Img2.kps,matches_inliers,None,**draw_params)
        
        plt.figure(1)
        plt.imshow(cv.cvtColor(img_inliers, cv.COLOR_BGR2RGB))
        plt.title("Inlier Matches for Total Selected Area")
        plt.axis('off')

        '''
        Stitch the Images
        '''
        img_stitch = simpleStitch(img1, img2, homo_mat)
        img_stitch_ = img_stitch.copy()
        cv.namedWindow("Point Selection", cv.WINDOW_NORMAL)
        cv.resizeWindow("Point Selection", 1960, 1080)
        cv.setMouseCallback('Point Selection',select_pts)
        while(1):
            cv.imshow('Point Selection',img_stitch_)
            if cv.waitKey(20) & 0xFF == 27:
                break
        cv.setMouseCallback('Point Selection',select_dst)
        while(1):
            cv.imshow('Point Selection',img_stitch_)
            if cv.waitKey(20) & 0xFF == 27:
                break
        pts = np.array(pts,dtype="float32")
        dst = np.array(dst,dtype="float32")
        cv.destroyAllWindows()
        if pts.shape[0] == 4 and dst.shape[0] ==4:
            img_stitch = transform.four_point_transform(img_stitch, pts, dst)
        else:
            img_stitch = img_stitch
        
        pts = []
        dst = []

        plt.figure(2)
        plt.imshow(cv.cvtColor(img_stitch, cv.COLOR_BGR2RGB))
        plt.axis('off')
        plt.show()

    '''
    Stitch the Arie Farm from lamp01 to lamp05
    
    Generate estimated homography matrix using matches between corresponding features in ROIs.
    Directly apply homography matrix to stitch all frames
    --------------------
    
    :param  img1_~img5_: original distorted images from each frame
    :type  img1_~img5_: nparray (768*1152*3)
    
    :param  homo_mat: homography matrix generated from feature estimation, using for image transformation
    :type  homo_mat: nparray (3*3)
    
    :return  img_stitch: stitching image of lamp14 to lamp18
    '''
    
    # # load the initial images and corresponding homo matrix
    # img1_ = cv.imread("dataset/office_farm/lamp_01.PNG")
    # img2_ = cv.imread("dataset/office_farm/lamp_02.PNG")
    # img3_ = cv.imread("dataset/office_farm/lamp_03.PNG")


    # # Define the operating lamp
    # lamp_id1 = 'lamp01'
    # lamp_id2 = 'lamp02'
    # lamp_id3 = 'lamp03'


    # # Enter the direction of the parameters
    # calib_dir = "/home/cxu-lely/kyle-xu001/Multi-Depth-Multi-Camera-Stitching/calib_params_office_farm"

    # # Generate the undistorted image according to intrinsic parameters
    # img1, _, _ = undistortion.undistort(img1_, lamp_id1, calib_dir)
    # img2, _, _ = undistortion.undistort(img2_, lamp_id2, calib_dir)
    # img3, _, _ = undistortion.undistort(img3_, lamp_id3, calib_dir)

    
    # '''Stitch Images'''    
    # # Param Tesing 2
    # homo_mat = np.array([[1.0016202444806064, -0.008750994107767975, 11.09472645657253,
    #                       0.022462672875005682, 0.9690859614898248, 332.5454427692357,
    #                       1.0605062766497904e-05, -2.8766106498432552e-05, 1.0]]).reshape(-1, 3)
    # img_stitch = ImageStitch.simpleStitch(img2, img3, homo_mat)

    # # homo_mat = np.array([[0.955153136287694, -0.036003672840083316, -10.840022964027122,
    # #                       -0.005812908840660401, 0.9446325649623217, 323.02301709013216,
    # #                       -3.3868807588141836e-05, -4.31029868417192e-05, 1.0]]).reshape(-1, 3)
    # # img_stitch = ImageStitch.simpleStitch(img3, img_stitch, homo_mat)

    # plt.figure(0)
    # plt.imshow(cv.cvtColor(img_stitch, cv.COLOR_BGR2RGB))
    # plt.axis('off')
    # plt.show()
    # # '''
    # Stitch the Right Area from lamp02 to lamp06
    
    # Generate estimated homography matrix using matches between corresponding features in ROIs.
    # Directly apply homography matrix to stitch all frames
    # --------------------
    
    # :param  img1_~img5_: original distorted images from each frame
    # :type  img1_~img5_: nparray (768*1152*3)
    
    # :param  homo_mat: homography matrix generated from feature estimation, using for image transformation
    # :type  homo_mat: nparray (3*3)
    
    # :return  img_stitch: stitching image of lamp14 to lamp18
    # '''
    
    # # load the initial images and corresponding homo matrix
    # img1_ = cv.imread("dataset/lamp_02_030610.PNG")
    # img2_ = cv.imread("dataset/lamp_03_122014.PNG")
    # img3_ = cv.imread("dataset/lamp_04_122014_for15.PNG")
    # img4_ = cv.imread("dataset/lamp_05_122014_for06.PNG")
    # img5_ = cv.imread("dataset/lamp_06_122014.PNG")

    # # Define the operating lamp
    # lamp_id1 = 'lamp02'
    # lamp_id2 = 'lamp03'
    # lamp_id3 = 'lamp04'
    # lamp_id4 = 'lamp05'
    # lamp_id5 = 'lamp06'

    # # Enter the direction of the parameters
    # calib_dir = "/home/cxu-lely/kyle-xu001/Multi-Depth-Multi-Camera-Stitching/calib_params_Mathe"

    # # Generate the undistorted image according to intrinsic parameters
    # img1, _, _ = undistortion.undistort(img1_, lamp_id1, calib_dir)
    # img2, _, _ = undistortion.undistort(img2_, lamp_id2, calib_dir)
    # img3, _, _ = undistortion.undistort(img3_, lamp_id3, calib_dir)
    # img4, _, _ = undistortion.undistort(img4_, lamp_id4, calib_dir)
    # img5, _, _ = undistortion.undistort(img5_, lamp_id5, calib_dir)
    
    # '''Stitch Images'''    
    # # Param Tesing 2
    # homo_mat = np.array([[0.9654716374414154, -0.037418169503540336, 8.454137475064552,
    #                       0.03041257452880801, 0.9411870423590862, 390.902621494968,
    #                       -0.00002238299370215597, -0.000028950718421592106, 1.0]]).reshape(-1, 3)
    # img_stitch = ImageStitch.simpleStitch(img1, img2, homo_mat)

    # homo_mat = np.array([[0.9909356479930977, -0.014386870228722386, 8.757529363659467,
    #                       0.06010029916698754, 0.9772712394834014, 824.234186668445,
    #                       -4.838670142009528e-06, -1.88189790991006e-05, 1.0]]).reshape(-1, 3)
    # img_stitch = ImageStitch.simpleStitch(img_stitch, img3, homo_mat)

    # homo_mat = np.array([[0.9517944903682141, -0.027727605854442557, 17.24844601329662,
    #                       0.0540712044523478, 0.911925755605298, 1234.9876293111838,
    #                       -3.166471839031637e-05, -2.9432142940408356e-05, 1.0]]).reshape(-1, 3)
    # img_stitch = ImageStitch.simpleStitch(img_stitch, img4, homo_mat)

    # homo_mat = np.array([[0.9040371566816637, -0.08038003595594359, 35.9401704232412,
    #                       0.030689953135582194, 0.7396604028039718, 1661.9534730902892,
    #                       -5.727569754952422e-05, -8.67077322367607e-05, 1.0]]).reshape(-1, 3)
    # img_stitch = ImageStitch.simpleStitch(img_stitch, img5, homo_mat)

    # plt.figure(0)
    # plt.imshow(cv.cvtColor(img_stitch, cv.COLOR_BGR2RGB))
    # plt.axis('off')
    # plt.show()
    
    # '''
    # Stitch the Right Area from lamp14 to lamp18
    
    # Generate estimated homography matrix using matches between corresponding features in ROIs.
    # Directly apply homography matrix to stitch all frames
    # --------------------
    
    # :param  img1_~img5_: original distorted images from each frame
    # :type  img1_~img5_: nparray (768*1152*3)
    
    # :param  homo_mat: homography matrix generated from feature estimation, using for image transformation
    # :type  homo_mat: nparray (3*3)
    
    # :return  img_stitch: stitching image of lamp14 to lamp18
    # '''
    
    # # load the initial images and corresponding homo matrix
    # img1_ = cv.imread("dataset/Paranoma/lamp_18_031513.PNG")
    # img2_ = cv.imread("dataset/Paranoma/lamp_17_031513.PNG")
    # img3_ = cv.imread("dataset/Paranoma/lamp_16_031513.PNG")
    # img4_ = cv.imread("dataset/Paranoma/lamp_15_031513.PNG")
    # img5_ = cv.imread("dataset/Paranoma/lamp_14_031513.PNG")

    # # Define the operating lamp
    # lamp_id1 = 'lamp18'
    # lamp_id2 = 'lamp17'
    # lamp_id3 = 'lamp16'
    # lamp_id4 = 'lamp15'
    # lamp_id5 = 'lamp14'

    # # Enter the direction of the parameters
    # calib_dir = "/home/cxu-lely/kyle-xu001/Multi-Depth-Multi-Camera-Stitching/calib_params_Mathe"

    # # Generate the undistorted image according to intrinsic parameters
    # img1, _, _ = undistortion.undistort(img1_, lamp_id1, calib_dir)
    # img2, _, _ = undistortion.undistort(img2_, lamp_id2, calib_dir)
    # img3, _, _ = undistortion.undistort(img3_, lamp_id3, calib_dir)
    # img4, _, _ = undistortion.undistort(img4_, lamp_id4, calib_dir)
    # img5, _, _ = undistortion.undistort(img5_, lamp_id5, calib_dir)

    # '''Stitch Images'''
    # # Param Tesing 1 
    # # homo_mat = np.array([[1.010638759789845, -0.020425372912002478, -14.695282380876835,
    # #                       0.0075814846672632484, 1.0066178703225623, 412.43074450438576,
    # #                       2.1888462791375407e-05, -2.2383079441019405e-05, 1.0]]).reshape(-1, 3)
    # # homo_mat = np.array([[
    #     # 0.9962750152836561, 0.011394294761734253, -15.863355184114155,
    #     # -0.005241890227618538, 1.0267777741157014, 411.4732629218854,
    #     # -0.000006746897821467397, 0.00004496214822227408, 1.0]]).reshape(-1, 3)
    # # img_stitch = ImageStitch.simpleStitch(img4, img5, homo_mat)

    # # homo_mat = np.array([[0.9058392172187458, 0.0004990937109045671, 47.304134684316686,
    # #                       -0.06666970815387915, 0.9191001811983562, 444.4072208488787,
    # #                       -0.00006927906859194502, -0.00004704323491312332,1.0]]).reshape(-1, 3)
    # # img_stitch = ImageStitch.simpleStitch(img3, img_stitch, homo_mat)

    # # homo_mat = np.array([[0.955857230597833, -0.057484217402180626, 71.90318975556875,
    # #                       -0.007353288099102101, 0.8925199803621107, 405.46235944750543,
    # #                       -0.00001106889426245175, -0.00008664024139811333, 1.0]]).reshape(-1, 3)
    # # img_stitch = ImageStitch.simpleStitch(img2, img_stitch, homo_mat)

    # # homo_mat = np.array([[1.10662515274729, -0.026531652925349116, -94.98865905700784,
    # #                       0.052902376594418724, 1.0362753243612786, 433.98218868249035,
    # #                       0.00008008349410686805, -0.000001116226659156957,1.0]]).reshape(-1, 3)
    # # img_stitch_ = ImageStitch.simpleStitch(img1, img_stitch, homo_mat)
    
    # # Param Tesing 2
    # homo_mat = np.array([[1.010638759789845, -0.020425372912002478, -14.695282380876835,
    #                       0.0075814846672632484, 1.0066178703225623, 412.43074450438576,
    #                       2.1888462791375407e-05, -2.2383079441019405e-05, 1.0]]).reshape(-1, 3)
    # img_stitch = ImageStitch.simpleStitch(img4, img5, homo_mat)

    # homo_mat = np.array([[0.9111138797319636, 0.025992907667642014, 43.51630224609544,
    #                       -0.08061531667103257, 0.976052629267486, 444.77675271224103,
    #                       -7.790215434632145e-05, 2.9319041438978976e-05, 1.0]]).reshape(-1, 3)
    # img_stitch = ImageStitch.simpleStitch(img3, img_stitch, homo_mat)

    # homo_mat = np.array([[0.9296608798773768, -0.04505056659216292, 78.06623318399255,
    #                       -0.016900992281276207, 0.917599731615178, 400.8510671764359,
    #                       -3.617106314861764e-05, -6.205541847742475e-05, 1.0]]).reshape(-1, 3)
    # img_stitch = ImageStitch.simpleStitch(img2, img_stitch, homo_mat)

    # homo_mat = np.array([[1.1386527445204575, 0.031927187738166204, -109.45549680535996,
    #                       0.04498832067998204, 1.0320031734670985, 463.08525965607436,
    #                       0.00010009854442945858, 0.00006315329701074331, 1.0 ]]).reshape(-1, 3)
    # img_stitch_ = ImageStitch.simpleStitch(img1, img_stitch, homo_mat)

    # plt.figure(0)
    # plt.imshow(cv.cvtColor(img_stitch_, cv.COLOR_BGR2RGB))
    # plt.axis('off')
    # plt.show()
    
    # '''Transform Stitching Result to avoid Distortion in final stitching'''
    # pts = np.array([[200, 0],
    #                 [1200, 0],
    #                 [800, 2000],
    #                 [500, 1900]])
    
    # dst = np.array([[180, 0],
    #                 [1250, 0],
    #                 [815, 1950],
    #                 [500, 1900]], dtype="float32")

    # img_stitch = transform.four_point_transform(img_stitch, pts, dst)

    # plt.figure(1)
    # plt.imshow(cv.cvtColor(img_stitch, cv.COLOR_BGR2RGB))
    # plt.axis('off')
    # plt.show()
    
    # pts = np.array([[200, 0],
    #                 [1200, 0],
    #                 [1200, 2150],
    #                 [200, 2150]])
    
    # dst = np.array([[180, 0],
    #                 [1100, 0],
    #                 [1100, 2150],
    #                 [200, 2150]], dtype="float32")

    # img_stitch = transform.four_point_transform(img_stitch, pts, dst)

    # plt.figure(1)
    # plt.imshow(cv.cvtColor(img_stitch, cv.COLOR_BGR2RGB))
    # plt.axis('off')
    # plt.show()


    # '''Using feature-based estimation to generate homography matrix'''
    # # load the matching images
    # img1_ = cv.imread("dataset/lamp1415161718-lamp19/lamp_19_031613.PNG")
    # lamp_id1 = 'lamp19'
    
    
    # img2 = img1
    # img1 = img_stitchq
    # cv.namedWindow("Area Selection", cv.WINDOW_NORMAL)
    # cv.resizeWindow("Area Selection", 800, 600)
    
    # ROIs1 = cv.selectROIs("Area Selection", img2)
    # ROIs2 = cv.selectROIs("Area Selection", img1)

    # for i in range(len(ROIs1)):
    #     ROIs1[i, 2] = ROIs1[i, 0] + ROIs1[i, 2]
    #     ROIs1[i, 3] = ROIs1[i, 1] + ROIs1[i, 3]
    #     ROIs2[i, 2] = ROIs2[i, 0] + ROIs2[i, 2]
    #     ROIs2[i, 3] = ROIs2[i, 1] + ROIs2[i, 3]

    # stitches = Stitch(img2, img1)
    # stitches.featureExtract(ROIs1, ROIs2)
    # homo_mat, matches_inliers = stitches.homoEstimate()
    # kps1 = stitches.Img1.kps
    # kps2 = stitches.Img2.kps

    # # Visualization
    # print("\nNumber of inlier matches: ", len(matches_inliers), "\n")
    # img_inliers = cv.drawMatches(
    #     img2, kps1, img1, kps2, matches_inliers, None, **draw_params)

    # plt.figure(2)
    # plt.imshow(cv.cvtColor(img_inliers, cv.COLOR_BGR2RGB))
    # plt.title("Inlier Matches for Total Selected Area [Num: %d]"%(len(matches_inliers)))
    # plt.axis('off')

    # '''Stitch the Images'''
    # img_stitch_ = ImageStitch.simpleStitch(img2, img1, homo_mat)

    # plt.figure(3)
    # plt.imshow(cv.cvtColor(img_stitch_, cv.COLOR_BGR2RGB))
    # plt.axis('off')
    # plt.show()

    '''
    Stitch the Left Area from lamp23 to lamp19
    
    Generate estimated homography matrix using matches between corresponding features in ROIs.
    Directly apply homography matrix to stitch all frames
    --------------------
    
    :param  img1_~img5_: original distorted images from each frame
    :type  img1_~img5_: nparray (768*1152*3)
    
    :param  homo_mat: homography matrix generated from feature estimation, using for image transformation
    :type  homo_mat: nparray (3*3)
    
    :return  img_stitch: stitching image of lamp23 to lamp19
    '''
    # load the initial images and corresponding homo matrix
    img1_ = cv.imread("dataset/Paranoma/lamp_23_031513.PNG")
    img2_ = cv.imread("dataset/Paranoma/lamp_22_031513.PNG")
    img3_ = cv.imread("dataset/Paranoma/lamp_21_031513.PNG")
    img4_ = cv.imread("dataset/Paranoma/lamp_20_031513.PNG")
    img5_ = cv.imread("dataset/Paranoma/lamp_19_031513.PNG")

    # Define the operating lamp
    lamp_id1 = 'lamp23'
    lamp_id2 = 'lamp22'
    lamp_id3 = 'lamp21'
    lamp_id4 = 'lamp20'
    lamp_id5 = 'lamp19'

    # Enter the direction of the parameters
    calib_dir = "/home/cxu-lely/kyle-xu001/Multi-Depth-Multi-Camera-Stitching/calib_params_Mathe"

    # Calculate the mapping matrix
    img1, _, _ = undistortion_mapping_test.undistort(img1_, lamp_id1, calib_dir)
    img2, _, _ = undistortion_mapping_test.undistort(img2_, lamp_id2, calib_dir)
    img3, _, _ = undistortion_mapping_test.undistort(img3_, lamp_id3, calib_dir)
    img4, _, _ = undistortion_mapping_test.undistort(img4_, lamp_id4, calib_dir)
    img5, _, _ = undistortion_mapping_test.undistort(img5_, lamp_id5, calib_dir)

    # Keep the img5 as background image, stitch from left to right
    img1 = cv.flip(img1, 0)
    img2 = cv.flip(img2, 0)
    img3 = cv.flip(img3, 0)
    img4 = cv.flip(img4, 0)
    img5 = cv.flip(img5, 0)

    '''Stitch Images'''
    homo_mat = np.array([[1.0223869691572842, -0.029757793305358708, -9.554931168189306,
                          -0.10257780654402092, 0.9829667304728431, 382.29502902382444,
                          2.9003442218944775e-05, -4.261237504410896e-05, 1.0]]).reshape(-1, 3)
    img_stitch = ImageStitch.simpleStitch(img2, img1, homo_mat)

    homo_mat = np.array([[0.895863627101159, -0.009314220498154528, 39.69811368863527,
                          -0.14319312890711713, 0.9238744611145023, 407.6464453689328, 
                          -0.00010585661979243849, 1.1089445174037917e-05, 0.9999999999999999]]).reshape(-1, 3)
    img_stitch = ImageStitch.simpleStitch(img3, img_stitch, homo_mat)

    homo_mat = np.array([[1.0093568603313154, -0.024777213229165367, -1.7110387731670804,
                          -0.0917809278675415, 0.9321125574704138, 497.27554411496135,
                          3.0226341371213186e-05, -4.5460503076694054e-05, 1.0]]).reshape(-1, 3)
    img_stitch = ImageStitch.simpleStitch(img4, img_stitch, homo_mat)

    homo_mat = np.array([[1.136892310007503, -0.030797753772207552, -25.882898617872147,
                          -0.08009097612575575, 1.0539222856379677, 474.91596299428886,
                          0.00012140360121091282, -3.315567108552388e-05, 1.0]]).reshape(-1, 3)
    img_stitch = ImageStitch.simpleStitch(img5, img_stitch, homo_mat)


    plt.figure(2)
    plt.imshow(cv.cvtColor(img_stitch, cv.COLOR_BGR2RGB))
    plt.axis('off')
    plt.show()

    # '''Transform Stitching Result to avoid Distortion in final stitching'''
    # width = img_stitch.shape[0]
    # pts = np.array([[270, 0],
    #                 [920, 0],
    #                 [930, width-910],
    #                 [180, width-400]])
    
    # dst = np.array([[270, 0],
    #                 [920, 0],
    #                 [910, width-775],
    #                 [250, width-775]], dtype="float32")

    # img_stitch = transform.four_point_transform(img_stitch, pts, dst)
    # img_stitch = cv.flip(img_stitch, 0)
    
    # plt.figure(3)
    # plt.imshow(cv.cvtColor(img_stitch, cv.COLOR_BGR2RGB))
    # plt.axis("off")
    # plt.show()

    # # '''Estimate the Homo Matrix for Final Stitching'''
    # # # load the matching images
    # # img1 = img_stitch
    # # img2 = img_stitch_
    
    # # # Define the size of opencv window
    # # cv.namedWindow("Area Selection", cv.WINDOW_NORMAL)
    # # cv.resizeWindow("Area Selection", 800, 600)
    
    # # ROIs1 = cv.selectROIs("Area Selection", img1)
    # # ROIs2 = cv.selectROIs("Area Selection", img2)

    # # for i in range(len(ROIs1)):
    # #     ROIs1[i, 2] = ROIs1[i, 0] + ROIs1[i, 2]
    # #     ROIs1[i, 3] = ROIs1[i, 1] + ROIs1[i, 3]
    # #     ROIs2[i, 2] = ROIs2[i, 0] + ROIs2[i, 2]
    # #     ROIs2[i, 3] = ROIs2[i, 1] + ROIs2[i, 3]

    # # stitches = Stitch(img1, img2)
    # # stitches.featureExtract(ROIs1, ROIs2)
    # # homo_mat, matches_inliers = stitches.homoEstimate()
    # # kps1 = stitches.Img1.kps
    # # kps2 = stitches.Img2.kps

    # # # Visualization
    # # print("\nNumber of inlier matches: ", len(matches_inliers), "\n")
    # # img_inliers = cv.drawMatches(
    # #     img1, kps1, img2, kps2, matches_inliers, None, **draw_params)

    # # plt.figure(4)
    # # plt.imshow(cv.cvtColor(img_inliers, cv.COLOR_BGR2RGB))
    # # plt.title("Inlier Matches for Total Selected Area [Num: %d]"%(len(matches_inliers)))
    # # plt.axis('off')

    # # '''Final Stitch'''
    # # img_stitch = ImageStitch.simpleStitch(img1, img2, homo_mat)
    
    
    # '''Final Stitch between two Area'''
    # homo_mat = np.array([[0.9131002758816589, -0.10864560677082075, 108.34106936850189, -0.22084159252494037, 0.7010361167342065, 2494.7795892747777, -3.353712513337784e-05, -0.00011457038122211423, 1.0]]).reshape(-1, 3)

    # #img_stitch = cv.flip(img_stitch, 0)
    # img_stitch = ImageStitch.simpleStitch(img_stitch, img_stitch_, homo_mat)
    # paranoma = img_stitch[500:5300,50:1450,:]
    
    # plt.figure(5)
    # plt.imshow(cv.cvtColor(np.rot90(paranoma), cv.COLOR_BGR2RGB))
    # plt.axis('off')
    # plt.show()


    '''
    Print the parameters of homography matrix
    '''
    np.set_printoptions(suppress=True)
    print(homo_mat.flatten().tolist())