Machine Learning with R Cookbook - Second Edition

How it works...

In this recipe, we demonstrate how to compress an image with SVD. In the first step, we use the package, bmp, to load the image, Lenna, to an R session. Then, as the read image is rotated, we can rotate the image back and use the plot function to plot Lenna in R (as shown in the figure in step 3). Next, we perform SVD on the image matrix to factorize the matrix. We then plot the percentage of variance explained in regard to the number of singular vectors.

Further, as we discover that we can use 18 components to explain 90 percent of the variance, we then use these 18 components to reconstruct Lenna. Thus, we make a function named lenna_compression with the purpose of reconstructing the image by matrix multiplication. As a ...