Menu
I'm trying to implement face detection in C#. I currently have a black + white outline of a photo with a face within it (Here). However i'm now trying to remove the noise and then dilate the image in order to improve reliability when i implement the detection.
The method I have so far is here:
using System;
using System.Collections.Generic
using System.ComponentModel;
using System.Data;
using System.Drawing;
using System.Linq;
using System.Text;
using System.Windows.Forms;
using System.Drawing.Imaging;
namespace ImageErosion
{
public partial class Form1 : Form
{
public int CompareEmptyColor { get; set; }
public Form1()
{
InitializeComponent();
}
private void btErodeImage_Click(object sender, EventArgs e)
{
Image inputImage = pbInputImage.Image;
Image result = Process(inputImage);
pbInputImage.Image = result;
}
unsafe public Image Process(Image input)
{
Bitmap bmp = (Bitmap)input;
Bitmap bmpSrc = (Bitmap)input;
BitmapData bmData = bmp.LockBits(new Rectangle(0, 0, bmp.Width, bmp.Height),
ImageLockMode.ReadWrite,
PixelFormat.Format1bppIndexed);
int stride = bmData.Stride;
int stride2 = bmData.Stride * 2;
IntPtr Scan0 = bmData.Scan0;
byte* p = (byte*)(void*)Scan0;
int nOffset = stride - bmp.Width * 3;
int nWidth = bmp.Width - 2;
int nHeight = bmp.Height - 2;
var w = bmp.Width;
var h = bmp.Height;
var rp = p;
var empty = CompareEmptyColor;
byte c, cm;
int i = 0;
// Erode every pixel
for (int y = 0; y < h; y++)
{
for (int x = 0; x < w; x += 3, i++)
{
// Middle pixel
cm = p[y * stride + x];
if (cm == empty) { continue; }
#region FirstRow
// Row 0
// Left pixel
if (x - 3 > 0 && y - 2 > 0)
{
c = p[(y - 2) * stride + (x - 3)];
if (c == empty) { continue; }
}
// Middle left pixel
if (x - 2 > 0 && y - 2 > 0)
{
c = p[(y - 2) * stride + (x - 2)];
if (c == empty) { continue; }
}
if (x - 1 > 0 && y - 2 > 0)
{
c = p[(y - 2) * stride + (x - 1)];
if (c == empty) { continue; }
}
if (y - 2 > 0)
{
c = p[(y - 2) * stride + x];
if (c == empty) { continue; }
}
if (x + 1 < w && y - 2 > 0)
{
c = p[(y - 2) * stride + (x + 1)];
if (c == empty) { continue; }
}
if (x + 2 < w && y - 2 > 0)
{
c = p[(y - 2) * stride + (x + 2)];
if (c == empty) { continue; }
}
if (x + 3 < w && y - 2 > 0)
{
c = p[(y - 2) * stride + (x + 3)];
if (c == empty) { continue; }
}
#endregion
#region SecondRow
// Row 1
// Left pixel
if (x - 3 > 0 && y - 1 > 0)
{
c = p[(y - 1) * stride + (x - 3)];
if (c == empty) { continue; }
}
if (x - 2 > 0 && y - 1 > 0)
{
c = p[(y - 1) * stride + (x - 2)];
if (c == empty) { continue; }
}
if (x - 1 > 0 && y - 1 > 0)
{
c = p[(y - 1) * stride + (x - 1)];
if (c == empty) { continue; }
}
if (y - 1 > 0)
{
c = p[(y - 1) * stride + x];
if (c == empty) { continue; }
}
if (x + 1 < w && y - 1 > 0)
{
c = p[(y - 1) * stride + (x + 1)];
if (c == empty) { continue; }
}
if (x + 2 < w && y - 1 > 0)
{
c = p[(y - 1) * stride + (x + 2)];
if (c == empty) { continue; }
}
if (x + 3 < w && y - 1 > 0)
{
c = p[(y - 1) * stride + (x + 3)];
if (c == empty) { continue; }
}
#endregion
#region ThirdRow
// Row 2
if (x - 3 > 0)
{
c = p[y * stride + (x - 3)];
if (c == empty) { continue; }
}
if (x - 2 > 0)
{
c = p[y * stride + (x - 2)];
if (c == empty) { continue; }
}
if (x - 1 > 0)
{
c = p[y * stride + (x - 1)];
if (c == empty) { continue; }
}
if (x + 1 < w)
{
c = p[y * stride + (x + 1)];
if (c == empty) { continue; }
}
if (x + 2 < w)
{
c = p[y * stride + (x + 2)];
if (c == empty) { continue; }
}
if (x + 3 < w)
{
c = p[y * stride + (x + 3)];
if (c == empty) { continue; }
}
#endregion
#region FourthRow
// Row 3
if (x - 3 > 0 && y + 1 < h)
{
c = p[(y + 1) * stride + (x - 3)];
if (c == empty) { continue; }
}
if (x - 2 > 0 && y + 1 < h)
{
c = p[(y + 1) * stride + (x - 2)];
if (c == empty) { continue; }
}
if (x - 1 > 0 && y + 1 < h)
{
c = p[(y + 1) * stride + (x - 1)];
if (c == empty) { continue; }
}
if (y + 1 < h)
{
c = p[(y + 1) * stride + x];
if (c == empty) { continue; }
}
if (x + 1 < w && y + 1 < h)
{
c = p[(y + 1) * stride + (x + 1)];
if (c == empty) { continue; }
}
if (x + 2 < w && y + 1 < h)
{
c = p[(y + 1) * stride + (x + 2)];
if (c == empty) { continue; }
}
if (x + 3 < w && y + 1 < h)
{
c = p[(y + 1) * stride + (x + 3)];
if (c == empty) { continue; }
}
#endregion
#region FifthRow
// Row 4
if (x - 3 > 0 && y + 2 < h)
{
c = p[(y + 2) * stride + (x - 3)];
if (c == empty) { continue; }
}
if (x - 2 > 0 && y + 2 < h)
{
c = p[(y + 2) * stride + (x - 2)];
if (c == empty) { continue; }
}
if (x - 1 > 0 && y + 2 < h)
{
c = p[(y + 2) * stride + (x - 1)];
if (c == empty) { continue; }
}
if (y + 2 < h)
{
c = p[(y + 2) * stride + x];
if (c == empty) { continue; }
}
if (x + 1 < w && y + 2 < h)
{
c = p[(y + 2) * stride + (x + 1)];
if (c == empty) { continue; }
}
if (x + 2 < w && y + 2 < h)
{
c = p[(y + 2) * stride + (x + 2)];
if (c == empty) { continue; }
}
if (x + 3 < w && y + 2 < h)
{
c = p[(y + 2) * stride + (x + 3)];
if (c == empty) { continue; }
}
#endregion
// If all neighboring pixels are processed
// it's clear that the current pixel is not a boundary pixel.
rp[i] = cm;
}
}
bmpSrc.UnlockBits(bmData);
return bmpSrc;
}
}
}
As I understand it, in order to erode the image (and remove the noise), we need to check each pixel to see if it's surrounding pixels are black, and if so, then it is a border pixel and we need not keep it, which i believe my code does, so it is beyond me why it doesn't work.
Any help or pointers would be greatly appreciated
Thanks, Chris
904
The OpenCV method is a common method in face detection. It firstly extracts the feature images into a large sample set by extracting the face Haar features in the image and then uses the AdaBoost algorithm as the face detector.
Face recognition is often described as a process that first involves four steps; they are: face detection, face alignment, feature extraction, and finally face recognition.
When it comes to a good, all-purpose face detector, I suggest using OpenCV's DNN face detector: It achieves a nice balance of speed and accuracy. As a deep learning-based detector, it's more accurate than its Haar cascade and HOG + Linear SVM counterparts. It's fast enough to run real-time on CPUs.
Face detection -- also called facial detection -- is an artificial intelligence (AI) based computer technology used to find and identify human faces in digital images.
A couple of bugaboos that jump out. The image format is 24bpp but you are reading bytes. That could sorta work if it is a pure black+white image but the left pixel would be at x - 3. Indexing x by 3 would also be wise.
Indexing the row is wrong, you multiply by w, you should multiply by stride.
179
you should take a look at the AForge.net Library ( http://code.google.com/p/aforge/ ). There are many different filters available for images. There you can also find examples how you can modify images directly
24
If you love us? You can donate to us via Paypal or buy me a coffee so we can maintain and grow! Thank you!
Donate Us With
