指纹领域最漂亮的公式

我觉得，这是指纹领域最漂亮而又非常简单的公式：
$$\theta =p\ast \text{atan2}(x-x_s,y-y_s).$$这里$p$代表奇异点的极性（+1或者-1），$(x_s,y_s)$代表奇异点的位置，$(x,y)$代表指纹图像中某点的坐标，$\theta$代表该点的方向或者相位。

这个公式既能用来建模指纹方向场的奇异点（singularity），又能用来表示指纹脊线的奇异点，也就是细节点（minutia）。

下面是生成两种方向场奇异点的MATLAB代码。

h = 30; w = 30; 
[X,Y] = meshgrid(1:w,1:h);

orient_continuous = -ones(h,w) * pi/2;

figure(1), quiver(X, Y, cos(orient_continuous), sin(orient_continuous),...
    'ShowArrowHead','off','AutoScaleFactor',0.7);
axis image

x_singular = w/2;
y_singular = h/2;

polarity = 1;
orient_singular = polarity * atan2(Y-y_singular, X-x_singular);
orient_combine = 0.5*mod(orient_continuous + orient_singular, 2*pi);

figure(2), quiver(X, Y, cos(orient_combine), sin(orient_combine),...
    'ShowArrowHead','off','AutoScaleFactor',0.7);
axis image

polarity = -1;
orient_singular = polarity * atan2(Y-y_singular, X-x_singular);
orient_combine = 0.5*mod(orient_continuous + orient_singular, 2*pi);

figure(3), quiver(X, Y, cos(orient_combine), sin(orient_combine),...
    'ShowArrowHead','off','AutoScaleFactor',0.7);
axis image

下面是生成两个朝向的指纹细节点（脊线奇异点）的MATLAB代码。

h = 500; w = 500; 
freq = 0.01;
[X,Y] = meshgrid(1:w,1:h);

phase_continuous = 2*pi*freq*Y;
figure(1), subplot(1,3,1), imshow(cos(phase_continuous),[]);

x_minu = w/2;
y_minu = h/2;

polarity = 1;
phase_minu = polarity * atan2(Y-y_minu, X-x_minu);
subplot(1,3,2), imshow(cos(phase_continuous+phase_minu),[]);

polarity = -1;
phase_minu = polarity * atan2(Y-y_minu, X-x_minu);
subplot(1,3,3), imshow(cos(phase_continuous+phase_minu),[]);