What Is Roof Edge Protection?

Edge recognition, also known as edge detection, is a process that mimics human vision. When detecting the edge of an object, it first performs rough detection on its contour points, and then connects the originally detected contour points through link rules. At the same time, it also detects and connects missing boundary points and removes false boundaries. The purpose of edge detection is to find information about the shape and reflection or transmittance in the image. It is one of the basic steps of image processing, image analysis, pattern recognition, computer vision, and human vision.

The edge of an image is one of the most basic and important features of an image.

Most of the information of the image exists in the edge of the image, which mainly manifests as the discontinuity of the local characteristics of the image, that is, the place where the gray level changes sharply in the image. Therefore, we define the edge as the area boundary where the gray level in the image changes drastically. According to the intensity of gray-scale changes, edges are usually divided into two types: step-shaped and roof-shaped. The gray value on both sides of the step edge changes significantly, while the roof edge is located at the junction of the increase and decrease of the gray value. Then, the first and second derivatives of step edge and roof edge can be used to represent the change of edge point. Therefore, for a step edge point, the first derivative of the gray change curve reaches a maximum value at this point, and the second derivative crosses zero at this point; for a roof edge point, the first order of the gray change curve The derivative crosses zero at this point; the second derivative reaches its maximum at this point. ^[1]

The essence of edge recognition is to use some algorithm to extract the boundary between the object in the image and the background. The change of image gray can be reflected by the gradient of image gray distribution, so we can use the local image differentiation technique to obtain the edge detection operator. The classic edge detection method is to construct an edge detection operator for a small neighborhood of pixels in the original image.

Roberts operator positioning is more accurate, but because it does not include smoothing, it is more sensitive to noise. The Prewitt operator and the Sobel operator are both first-order differential operators. The former is an average filter, and the latter is a weighted average filter. The detected image edges may be larger than 2 pixels. Both of them have better detection effect on gray-gradient and low-noise images, but the processing effect is not ideal for images with many complex noises. The LOG filter method determines the edge point by detecting the second derivative zero-crossing point. The in the LOG filter is proportional to the width of the low-pass filter. The larger the , the more significant the smoothing effect is, and the better the noise is removed, but the greater the loss of image detail, the lower the edge accuracy. Therefore, there is a contradiction between the edge positioning accuracy and the noise level elimination, and should be appropriately selected according to the specific requirements for noise level and edge point positioning accuracy requirements. Moreover, the LOG method does not solve the specific method of how to organize the edge maps output by different scale filters into a single, correct edge map. Canny's method uses the first derivative to determine the edge points. It is one of the best operators for detecting stepped edges in traditional first-order differentiation. It has stronger denoising ability than the minimum algorithm of Roberts operator, Sobel operator and Prewitt operator, but it is also easy to smooth out some edge information. ^[2]