Programming with the Kinect™ for Windows® Software Development Kit by David Catuhe

Author:David Catuhe [David Catuhe]
Language: eng
Format: epub
Tags: COMPUTERS / Software Development & Engineering / General
ISBN: 9780735667808
Publisher: Microsoft Press
Published: 2012-09-12T16:00:00+00:00

Creating a learning machine

You now have a good pattern-matching algorithm and a tool to normalize your gestures. You are ready to create the learning machine that will aggregate these two tools. The first step is to create a class to represent the learning machine database.

The RecordedPath class

As explained previously, the learning machine is a database filled with templates of recorded gestures. To represent a template, you can use the following class:

using System; using System.Collections.Generic; using System.Diagnostics; using System.Windows; using Kinect.Toolbox.Gestures.Learning_Machine; // For the GoldenSection methods using System.Windows.Media.Imaging; using System.Windows.Media; namespace Kinect.Toolbox { [Serializable] public class RecordedPath { List<Vector2> points; readonly int samplesCount; [NonSerialized] WriteableBitmap displayBitmap; public List<Vector2> Points { get { return points; } set { points = value; } } public WriteableBitmap DisplayBitmap { get { if (displayBitmap == null) { displayBitmap = new WriteableBitmap(200, 140, 96.0, 96.0, PixelFormats.Bgra32, null); byte[] buffer = new byte[displayBitmap.PixelWidth * displayBitmap.PixelHeight * 4]; foreach (Vector2 point in points) { int scaleX = (int)((point.X + 0.5f) * displayBitmap.PixelWidth); int scaleY = (int)((point.Y + 0.5f) * displayBitmap.PixelHeight); for (int x = scaleX - 2; x <= scaleX + 2; x++) { for (int y = scaleY - 2; y <= scaleY + 2; y++) { int clipX = Math.Max(0, Math.Min(displayBitmap.PixelWidth - 1, x)); int clipY = Math.Max(0, Math.Min(displayBitmap.PixelHeight - 1, y)); int index = (clipX + clipY * displayBitmap.PixelWidth) * 4; buffer[index] = 255; buffer[index + 1] = 0; buffer[index + 2] = 0; buffer[index + 3] = 255; } } } displayBitmap.Lock(); int stride = displayBitmap.PixelWidth * displayBitmap.Format.BitsPerPixel / 8; Int32Rect dirtyRect = new Int32Rect(0, 0, displayBitmap.PixelWidth, displayBitmap.PixelHeight); displayBitmap.WritePixels(dirtyRect, buffer, stride, 0); displayBitmap.AddDirtyRect(dirtyRect); displayBitmap.Unlock(); } return displayBitmap; } } public RecordedPath(int samplesCount) { this.samplesCount = samplesCount; points = new List<Vector2>(); } public void CloseAndPrepare() { points = GoldenSection.Pack(points, samplesCount); } public bool Match(List<Vector2> positions, float threshold, float minimalScore, float minSize) { if (positions.Count < samplesCount) return false; if (!positions.IsLargeEnough(minSize)) return false; List<Vector2> locals = GoldenSection.Pack(positions, samplesCount); float score = GoldenSection.Search(locals, points, -MathHelper.PiOver4, MathHelper.PiOver4, threshold); Debug.WriteLine(score); return score > minimalScore; } } }

The RecordedPath class provides the following information:

A list of Vector2 (the template itself)

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.