Registering Hand-Held Pictures

发表史蒂夫·埃丁（Steve Eddins），，，，December 23, 2020

18views (last 30 days) |0Likes|0comment

            The typical modern French* horn, pictured below, has about 23 total feet of tubing. At the beginning and the end, the tubing is conical. In the middle, the tubing is cylindrical.
           

            Depending on which valve levers are pressed, a player might be buzzing into single tube that is anywhere from 9 feet to 17 feet long (approximately). For a personal project, I wanted to create a visual illustration of these various lengths. I used an extension pole and got some help in taking several pictures. My helper used a handheld phone.
           

              A = imread('a.jpg'）；
             

              b = imread（'B.jpg'）；
             

              C = imread('C.jpg'）；
             

              Tiledlayout（2,2）
             

              Nexttile
             

              imshow(A,“插值”，，，，“双线”）
             

              title(“图像A（f号长，12英尺）”）
             

              Nexttile
             

              imshow(B,“插值”，，，，“双线”）
             

              title(“图像b（b-flat喇叭的长度，9英尺）”）
             

              Nexttile
             

              imshow(C,“插值”，，，，“双线”）
             

              title('Image C (length of B horn, 17 feet)'）
             

            我想使用这些图片制作一个复合材料，使您可以在视觉上比较三极长。问题在于，这三张图片不彼此对齐，这使得创造了精确的复合挑战。
           

            Let me use
            imshowpair
            to illustrate what I mean, using images A and B.
           

Clf

              imshowpair(A,B)
             

            看来这两个图像通过翻译和较小的旋转而彼此不同。
           

            为了实现我的目标，我需要首先让这三张图片彼此对齐。然后，我可以裁剪它们并创建我的复合材料。
           

            Sounds like an image registration and spatial referencing problem!
           

            Here's the plan:
           

Infer a geometric transform that aligns image B to image A.
Infer a geometric transform that aligns image C to image A.
使用上面发现的转换，将图像B和C翘曲到与图像A相同的空间参考。
Crop all three images closely around me holding the pole in the center.
Create a composite of the three images.

            我将在图像处理工具箱和计算机视觉工具箱中使用工具。
           

Register image B to image A

            将图像转换为灰度以进行注册程序。
           

              ag = rgb2gray（a）;
             

              bg = rgb2gray（b）;
             

            在这两个图像中使用冲浪功能检测（来自计算机视觉工具箱）。
           

              ％检测冲浪功能
             

              A_points = detectSURFFeatures(Ag);
             

              b_points =检测表（bg）;
             

            从检测到的特征点提取功能信息。
           

              [a_features，a_valid_points] = extractfeatures（ag，a_points）;
             

              [B_features,B_valid_points] = extractFeatures(Bg,B_points);
             

            匹配两个图像的功能点。
           

              index_pairs_b_a = matchfeatures（a_features，b_features）;
             

              a_matched_points = a_valid_points（index_pairs_b_a（：，1））;
             

              B_matched_points = B_valid_points(index_pairs_B_A(:,2));
             

            Infer a projective geometric transformation that will align image B with image A.
           

              tform_B_A = estimateGeometricTransform(B_matched_points,A_matched_points,'projective'）；
             

              tform_b_a.t
             

                  ans =
                  3×3single matrix
                 

                  1。02110。03460。0000 -0.0196 1.0139 0.0000 31.3027 47.2352 1.0000

            Align image B to image A. Note especially the use of the
            输出视图
            选项
            IMWARP
            ;这用于计算与图像A所占据的相同空间矩形的图像翘曲结果。
           

              a_ref = imref2d（size（ag））;
             

              B_reg = imwarp(B, tform_B_A,'OutputView'，a_ref）;
             

            视觉检查结果。
           

              figure
             

              imshowpair(A,B_reg,“插值”，，，，“双线”）
             

注册图像c使用相同的步骤图像A图像A

              Cg = rgb2gray(C);
             

              C_points = detectSURFFeatures(Cg);
             

              [C_FEATURES，C_VALID_POINTS] = ExtractFeatures（cg，c_points）;
             

              index_pairs_c_a = matchfeatures（a_features，c_features）;
             

              a_matched_points = a_valid_points（index_pairs_c_a（：，1））;
             

              c_matched_points = c_valid_points（index_pairs_c_a（：，2））;
             

              tform_C_A = estimateGeometricTransform (C_matched_points,A_matched_points,'projective'）；
             

              c_reg = imwarp（c，tform_c_a，'OutputView'，a_ref）;
             

              imshowpair(A,C_reg,“插值”，，，，“双线”）
             

Crop images to a common rectangle

            既然图像A，B和C都已注册为图像A的空间参考，则可以使用相同位置的相同矩形裁剪所有三个。
           

              r = [1100 450 600 1600];
             

              A_c = imcrop(A,r);
             

              b_reg_c = imcrop（b_reg，r）;
             

              c_reg_c = imcrop（c_reg，r）;
             

Make a tiled composite

            最后，我会用
            imtile
            为了制作可以保存到JPEG文件的图块复合图像。
           

              ABC = imtile({B_reg_c,A_c,C_reg_c},。。。
             

              'BorderSize'，50，。。。
             

              'BackgroundColor'，，，，'白色的'）；
             

              imshow(ABC,“插值”，，，，“双线”）
             

            令我惊讶的是，我可以在腿上握住的乐器确实可以是一根管子，只要右图所示的杆子。
           

            *现代乐器不是法国的，所以
            法国号角
            is a misnomer, strictly speaking. Many players refer to the instrument simply as a
            喇叭，，，，
            and this is the recommendation of the International Horn Society. In the United States,however, most people who are not orchestral musicians will be confused by the term
            喇叭
            ，，，，and so I use
            法国号角
            here.
           