How will the blur work on the edge cases? Will the kernel be applied with the current pixel be on the corner or the edge instead of in the middle? Or we will just cut the kernel into smaller ones?
yangbright-2001
Convolution operation is the core of convolution-based image processing. The kernel is slid over every pixel of the image. At each position, a weighted sum of the kernel values and the pixel values that fall under the kernel is calculated. This sum then replaces the pixel value at the position of the center of the kernel.
I also search some complementary info regarding edge handling. As the kernel moves over the edges of the image, parts of the kernel may extend beyond the image boundaries. Different strategies can be used to handle this, such as padding the image with zeros (zero-padding), replicating the edge values, or wrapping the image around.
How will the blur work on the edge cases? Will the kernel be applied with the current pixel be on the corner or the edge instead of in the middle? Or we will just cut the kernel into smaller ones?
Convolution operation is the core of convolution-based image processing. The kernel is slid over every pixel of the image. At each position, a weighted sum of the kernel values and the pixel values that fall under the kernel is calculated. This sum then replaces the pixel value at the position of the center of the kernel. I also search some complementary info regarding edge handling. As the kernel moves over the edges of the image, parts of the kernel may extend beyond the image boundaries. Different strategies can be used to handle this, such as padding the image with zeros (zero-padding), replicating the edge values, or wrapping the image around.