003_中英文对照版_.12.2.ppt

Example 3.9 An alternate definition of the digital second derivatives takes into account diagonal elements , and can be implemented using the mask: 另一种离散化的二阶导数定义，考虑了对角元素。该定义可用如下掩码实现： * (a) Image of the North Pole of the moon (b) L-filtered image, uint8 formats. f=imread(moon.tif); imshow(f) w = [0 1 0;1 -4 1;0 1 0]; g1=imfilter(f,w,replicate); imshow(g1,[]); Figure 3.16 (P102) * Example 3.9 All pixels of the image in Figure 3.16(b) are positive. Because of the negative center filter coefficient, we know that we can expect in general to have a Laplacian image with negative values . 图3.16(b)图像中的所有像素值都是正的。因为滤波器中心系数是负的，我们可以预料到经过拉普拉斯变换后的结果图像，其像素值都是负的。 However, f in this case is of class uint8 and , as discussed in the previous section, filtering with imfilter gives an output that is of the same class as the input image, so negative values are truncated. 然而，此例中f是uint8类型的。正如前面讨论过的，用函数进行滤波后得到的输出结果图像与输入图像的类型是一样的，因而负值就会被裁剪掉。 We get around this difficulty by converting f to class double before filtering it: 为了回避这一难题，在滤波之前，我们将图像f的类型转换成double类型。 f2=im2double(f); g2=imfilter(f2,w,replicate); * f2=im2double(f); g2=imfilter(f2,w,replicate); imshow(g2,[]); g = f2-g2; imshow(g) (c) L-filtered, double formats (d) obtained by (a)-(c) Figure 3.16 (P102) * Example 3.9 The result ,shown in Fig.3.16(c), is more what a properly processed Lanlacian image should look like. 经拉普拉斯变换后的正确的结果图像，看起来应像图3.16(c)所示的那样。 Finally, we restore the gray tones lost by using the Laplacian by subtracting (because the center coefficient is negative) the Laplacian image from the original image: 最后，通过将原图像减去拉普拉斯变换后的结果图像，我们就将在拉普拉斯变换后所丢失的色调进行恢复。 g = f2-g2; imshow(g) The result ,shown in Fig.3.16(d), is sharper than the original image. 恢复结果如图3.16(d)所示，和原图像比起来，得到了锐化。 * (c) L-filtered, double formats (d) obtained by (a)-(c) Figure 3.16 (P102) * Example 3.10 (P103) implementation: f=imread(moon.tif); %1 w4=fspeci