In this example, we compare the processing speed of convolution in the spatial domain to that of the same in the frequency domain. Consider a gray scale image of size 512 * 512 pixels. We wish to sharpen the image by processing it with a FIR high-pass filter of size 11 * 11. For simplicity, let’s ignore any symmetrical properties of the filter as it can be equally taken advantage of in both direct convolution and FFT-based convolution. Determine which approach will be faster, block convolution using overlap–add method with 64-point FFT or direct convolution in spatial domain?