Compass Masks in Image Processing

The Sobel and Prewitt edge detection masks are not rotationally symmetric or isotropic, which means edges at all orientations contribute to the result. This is different from the Laplacian masks, which are rotationally symmetric.

The sign of the result (positive or negative) from applying a Laplacian mask tells us what side of the edge is brighter.

The larger the sum of the coefficients in a convolution mask, the less the processed image is changed from the original image.

<p>The two metrics that can be determined are the edge magnitude and the edge direction.</p> Signup and view all the answers

<p>Unlike the Sobel and Prewitt edge detection masks, the Laplacian masks are rotationally symmetric or isotropic, which means edges at all orientations contribute to the result.</p> Signup and view all the answers

<p>If the sum of the coefficients in a convolution mask is zero, it indicates that the mask is only interested in edge information, and not in retaining most of the information from the original image.</p> Signup and view all the answers

<p>The resulting pixel value will depend most heavily upon the current value, with only minimal contribution from the surrounding pixel values.</p> Signup and view all the answers

<p>They are called compass masks because they are defined by taking a single mask and rotating it to the eight major compass orientations: North, Northwest, West, Southwest, South, Southeast, East, and Northeast.</p> Signup and view all the answers

<p>The edge magnitude is defined as the maximum value found by the convolution of each of the masks with the image.</p> Signup and view all the answers

<p>The edge direction is defined by the mask that produces the maximum magnitude.</p> Signup and view all the answers

<p>The Robinson compass masks are easier to implement, as they rely only on coefficients of 0, 1, and 2, and are symmetrical about their directional axis.</p> Signup and view all the answers

<p>We only need to compute the results on four of the masks; the results from the other four can be obtained by negating the results from the first four.</p> Signup and view all the answers

<p>The formula is: <code>pixel = abs(p1-p4)+abs(p2-p3)</code>, which is then clamped to the 0-255 range.</p> Signup and view all the answers

<p>The final formula uses the horizontal and vertical components in the form <code>pixel = SQRT((X*X)+(Y*Y))</code>, where <code>X = abs(p1-p4)</code> and <code>Y = abs(p2-p3)</code>.</p> Signup and view all the answers

<p>The Roberts Edge detector is fast but subject to interference by noise, and if edges are not very sharp, the filter will tend not to detect the edge.</p> Signup and view all the answers

<p>The Sobel operator uses a row mask and a column mask to approximate the gradient in the horizontal and vertical directions. The results from these two masks are then combined to compute the edge magnitude and direction.</p> Signup and view all the answers

<p>The two metrics calculated are the edge magnitude and the edge direction.</p> Signup and view all the answers

<p>The Prewitt operator is similar to the Sobel, but with different mask coefficients.</p> Signup and view all the answers