When I down-sample an image 4 time or higher using image.scale, significant aliasing artifacts will appear, even using bicubic interpolation algorithm. I guess that is properly due to the fact that image.scale does not blur the image before down-sampling. It might be good if it can blur the image first using Gaussian filter or the cubic filter (which is the default blur kernel of imresize function in Matlab).
When I down-sample an image 4 time or higher using image.scale, significant aliasing artifacts will appear, even using bicubic interpolation algorithm. I guess that is properly due to the fact that image.scale does not blur the image before down-sampling. It might be good if it can blur the image first using Gaussian filter or the cubic filter (which is the default blur kernel of imresize function in Matlab).