## 個人檔案

###### 關於

There are many ways to detect edges in an image, from the simplest to the most sophisticated. Simple Edge Detection This simple approach uses the default threshold in our Image Processing Toolbox. This algorithm works when the lightest pixel in the image is strictly brighter than the darkest pixel. Edge Detection through Convolution An example of a simple edge detector is the Sobel operator. A Sobel operator is the convolution of the image with the filter kernel, or the convolution of the image with a derivative of the image. In our example, we convolve the image with the filter kernel to determine the horizontal and vertical derivatives. From the derivatives, the image is thresholded to create a strong edge. Edge Detection using Canny Canny edge detector is the most general-purpose of these 5 algorithms because it is the best at detecting a variety of edge types. However, this algorithm is also quite slow. In our example, we run the image through the Canny edge detector. We create 3 binary images, one for each Canny edge detector location. From these 3 binary images, we select the largest component. Note: If there are multiple edge pixels in a single location, the resulting number of edges will be the size of the largest component. This is true even for multiple edge pixels that are close together. Edge Detection Using Sobel Here we use the Sobel operator to detect edges in the image. We use the Sobel operator because it's a fast, low-level algorithm. Edge Detection Using SobelX The SobelX algorithm is similar to the Sobel but is symmetric for detecting edges. It is equivalent to the Sobel but works much faster. In our example, we use the SobelX operator. Edge Detection Using Laplacian The Laplacian operator is another very fast algorithm for detecting edges. The idea behind this approach is to analyze the first derivatives of the image (forward and backward). We do this by going through the image row-by-row and adding a new pixel to the image each time we find a gradient of the image. Edge Detection Using "Gaussian" This algorithm finds a maximum response location in the image. The algorithm assumes the image is not too smooth, so the "Gaussian" is used as a smoothing kernel a5204a7ec7

What the user interface does: You can select one, two, or three adjacent edges to be made sharp or blurred. The sharpening effect or blur is uniform and therefore applies equally to the entire edge. Note that you cannot choose edges which are outside the region you are working with. You must first do a Select Edges by Point Selection. The Blur is limited to the area that is currently selected. The filter changes the selected edges from solid to dotted. The filter effects the entire image, not just the selected edge When the document is saved, the Sharpening filter is added to the image processing list. What the images show: The five methods are: Select Edges Fast Select Edges Bilateral Select Edges Adaptive Select Edges Smoothing Select Edges Uniform The five methods are described as follows: Select Edges Fast Noise, grain and linear Select Edges Bilateral Noise, grain and linear Select Edges Adaptive Noise, grain and linear Select Edges Smoothing Noise, linear Select Edges Uniform Noise, grain and linear Select Edges Fast Displays all edges, plus one with the color of the background. You can sharpen all edges except the background. Select Edges Fast ignores the background color. Select Edges Fast is useful when you do not want to sharpen the background. Select Edges Bilateral Displays all edges, plus two with alternating color. You can sharpen or blur only one edge, and sharpen or blur the other edge at the same time. Select Edges Bilateral ignores the background color. Select Edges Bilateral is useful when you do not want to sharpen the background. Select Edges Adaptive Displays all edges, plus three with alternating color. Selects edges based on the strength of the edge in relation to the background. Select Edges Adaptive is useful when you do not want to sharpen the edges. Select Edges Smoothing Displays all edges, plus five with alternating color. Selects edges based on the strength of the edge in relation to the background. Select Edges Smoothing is useful when you want to sharpen