@ai-on-browser/data-analysis-models/lib/model/log.js

Data analysis model package without any dependencies

/**
 * Laplacian of gaussian filter
 */
export default class LoG {
	// https://algorithm.joho.info/image-processing/laplacian-of-gaussian-filter/
	// https://betashort-lab.com/%E7%94%BB%E5%83%8F%E5%87%A6%E7%90%86/log%E3%83%95%E3%82%A3%E3%83%AB%E3%82%BF/
	/**
	 * @param {number} th Threshold
	 */
	constructor(th) {
		this._threshold = th
		this._k = 5
		this._s = 3
	}

	_convolute(x, kernel) {
		const a = []
		for (let i = 0; i < x.length; i++) {
			a[i] = []
			for (let j = 0; j < x[i].length; j++) {
				let v = 0
				for (let s = 0; s < kernel.length; s++) {
					let n = i + s - Math.floor(kernel.length / 2)
					n = Math.max(0, Math.min(x.length - 1, n))
					for (let t = 0; t < kernel[s].length; t++) {
						let m = j + t - Math.floor(kernel[s].length / 2)
						m = Math.max(0, Math.min(x[n].length - 1, m))
						v += x[n][m] * kernel[s][t]
					}
				}
				a[i][j] = v
			}
		}
		return a
	}

	/**
	 * Returns predicted edge flags.
	 * @param {Array<Array<number>>} x Training data
	 * @returns {Array<Array<boolean>>} Predicted values. `true` if a pixel is edge.
	 */
	predict(x) {
		const k = [
			[0, 0, 1, 0, 0],
			[0, 1, 2, 1, 0],
			[1, 2, -16, 2, 1],
			[0, 1, 2, 1, 0],
			[0, 0, 1, 0, 0],
		]
		const gl = this._convolute(x, k)

		const g = []
		for (let i = 0; i < gl.length; i++) {
			g[i] = []
			for (let j = 0; j < gl[i].length; j++) {
				g[i][j] = gl[i][j] > this._threshold
			}
		}
		return g
	}
}