点,线,形的检测.ppt

点、线、形的检测 Detect / Enhance Dots Laplacian filter is a good detector for dots (and lines) Detect / Enhance Dots Isolation point detector Edge Detector Edge detector (with directivity) Optimal Edge Detection Create 8-direction edge detector Detect / Enhance Lines Thin line detector Defined Pattern Enhancement Find shape that is pattern defined Image Matching How to find out given template image from a bigger image Image Matching Image Matching How to find out given small template image from a bigger image Normalization (Advanced Matching) To avoid effect of gray-values in the template and original image detected 几何变换 G = T( F ) Input image F (x,y); Output image G (u,v) G = T( F ) u = Tu(x, y) v = Tv(x, y) 平移Shift / 放缩Resize u=kx x + cx v=ky y + cy 旋转 Rotation u = cosqx - sinqy v = sinqx + cosqy 切变 (shear) 透视 Perspective 波浪 Wave polynomial transform u = S kijxiyj; v = S kijxiyj 极坐标 polar coordinates 球面 sphere Problem u = Tu(x, y) v = Tv(x, y) Nearest point 2. Linear method Render a output image pixel by pixel / line by line For each pixel in output image, we using inverse function T’(u,v) to evaluate (x,y) Unknown transform functionbut we have known his results of typical points A a B b C c : (xA,yA) (xa,ya) Linear functions xa = k1xA+k2yA + k3 xb = k1xB+k2yB + k3 xc = k1xC+k2yC + k3 ya = k4xA+k5yA + k6 yb = k4xB+k5yB + k6 yc = k4xC+k5yC + k6 Apply following functions in the trigon u = k1x+k2y + k3 v = k4x+k5y + k6 Fill all of image 求解 k1,k2,k3 k4,k5,k6 * * Detection for dot,line and shape 中南大学信息科学与工程学院 0 1 0 1 - 4 1 0 1 0 For black dots on white ground 0 -1 0 -1 4 -1 0 -1 0 For white dots on black ground 0 -1 0 -1 4 -1 0 -1 0 Small region point detector -1 -2 -1 -2 12 -2 -1 -2 -1 1 1 1 1 1 1 -1 -1 -1 1 1 -1 -1 -1 1 1 -1 -1 -1 1 1 1 1 1 1 -1 0 1 -1 0 1 -1 0 1 -1