CANNY EDGE DETECTION

Canny edge detection 的步驟：

1. 將原始影像和高斯濾波器做摺積。

2. 利用一次微分的遮罩對影像每一個像素求得四個方向的邊線強度

3. 找出四個方向中的最大值作為目前像素的邊線強度。

4. 依據步驟 3 將梯度方向分成四個區域

5. 非最大值刪除：沿著梯度方向找出最大值，並將其保留，其餘均設為零。

6. 設定兩個閥值 low T 和 high T ，用 high T 挑選出最佳的邊線像素，再從鄰近

點（neighbor)中找出梯度強度大於 low T 的像素，即可得到我們要的邊緣輪廓。

UML處可以選取影像路徑-Canny edge detection steps:

An original image and the Gaussian filter to do the convolution.

Edge intensity of 2 using the first derivative of the mask obtained for each pixel on the image in four directions

Find the maximum value in the four directions as the edge intensity of the pixel.

(4) in accordance with step 3 of the direction of the gradient is divided into four regions

5 non-maximal Delete: to find out the maximum value along the direction of the gradient, and reserves the right, the rest are set to zero.

6 set two thresholds a low T and high T-high T to select the best edge pixels, from the neighboring

Points (The neighbor) to find the gradient strength is greater than the pixels of a low T, you can get to the edge contour.

UML at you can select the image path