As a reminder, convolution works by centering the kernel over an image pixel and replacing that pixel’s value with a weighted sum of itself and surrounding pixels, using the filter values as weights, and repeating this process for every pixel in the image.