obartra/ssim

import { conv2 } from './conv2'
import { Matrix, Shape } from '../types'

/**
 * Rotates a matrix 180deg.
 *
 * @example
 * 1 2 3 4  becomes:  8 7 6 5
 * 5 6 7 8            4 3 2 1
 *
 * @method rotate1802d
 * @param {Matrix} mx - The input matrix
 * @returns {Matrix} out - The rotated matrix
 * @private
 * @memberOf matlab
 * @since 0.0.2
 */
function rotate1802d({ data: ref, width, height }: Matrix): Matrix {
  const data = new Array(ref.length)

  for (let i = 0; i < height; i++) {
    for (let j = 0; j < width; j++) {
      data[i * width + j] = ref[(height - 1 - i) * width + width - 1 - j]
    }
  }

  return {
    data,
    width,
    height,
  }
}

/**
 * Given a matrix X and a two-dimensional FIR filter h, filter2 rotates your filter matrix 180
 * degrees to create a convolution kernel. It then calls conv2, the two-dimensional convolution
 * function, to implement the filtering operation.
 *
 * This method mimics Matlab's `filter2` method
 *
 * @method filter2
 * @param {Matrix} h - The FIR filter
 * @param {Matrix} X - The input matrix
 * @param string [shape='same'] - The convolution shape
 * @returns {Matrix} conv - The 2D convolution of X with h
 * @public
 * @memberOf matlab
 * @since 0.0.2
 */
export function filter2(h: Matrix, X: Matrix, shape: Shape = 'same'): Matrix {
  return conv2(X, rotate1802d(h), shape)
}