A filter (also known as a kernel) is a small matrix used in convolution operations.

Convolution filters are commonly used in image processing to modify images or extract features.

The dimensions of a convolution filter are typically small, odd, and square. For example, convolution filters are typically \(3 \times 3\) or \(5 \times 5\) matrices. Odd dimensions are preferred to even dimensions.