Image Processing: Skin Detection, Some Filters and EXIF Tags

, 17 Jul 2009 CPOL
 Rate this:
Simple algorithms of skin detection and some useful filters

Introduction

RedMatter library is simple tool, which includes filters for skin detection and color correction and instruments for EXIF tags reading. Also you may develop your own filter and use it - it's very easy. Now RedMatter includes the following filters and tools:

• Skin detection filters: implementation of apriority and parametric algorithms
• Soft contour filter
• Rectangle vignette filter
• Color wave filter
• Selective monochromatic filter
• Statistic color correction tool

Using the Code

Skin Detection Filters

Apriority skin detection methods use a set of rules. These rules create a polygon in the color space. For example:

```(R > 95 and G > 40 and B < 20 and
max{R, G, B} – min{ R, G, B }> 15
and |R - G| > 15 and R > G and R > B)
or
(R > 220 and G > 210 and B > 170 and
|R - G| = 15 and
R > B and G > B)```

Or:

All skin detection classes implement a simple interface `ISkinDetector`:

```public interface ISkinDetector: IImageSelector
{
bool IsSkin(Color color);
}   ```

You may use these filters very easily:

```    BaseDetector detector = new BaseDetector(sourceImage, new SimpleSkinDetector());
pbxImage.Image = detector.SkinDetectionImage; ```

The library includes six apriority filters with different sets of rules.

Figure 1. Apriority skin detection.

Parametric methods (for example, the Gauss method) use the comparison image with the sample of skin. Skin color distribution can be modelled by an elliptical Gaussian joint probability density function (PDF), de?ned as:

Here, c is a color vector and µ and Ss are the distribution parameters.

where n is the total number of skin color samples cj.

Example of usage:

```// corrImage – It is sample of skin.
GaussianSkinDetector gauss = new GaussianSkinDetector(corrImage);
gauss.Threshold = 6;
BaseDetector detector = new BaseDetector(sourceImage, gauss);
pbxImage.Image = detector.SkinDetectionImage;```

Soft Contour Filter

Example of usage:

```SoftContourFilter soft = new SoftContourFilter(sourceImage);
soft.Softness = 20.0;
soft.Threshold = 125;
pbxImage.Image = soft.ResultImage;```

Figure 2. Soft contour filter.

Rectangle Vignette Filter

Example of code:

```VignetteFilter vign = new VignetteFilter(sourceImage);
vign.VignetteWidth = 50;
pbxImage.Image = vign.ResultImage;```

Figure 3. Simple vignette.

Color Wave Filter

This effect like the noise on the old TV.

Figure 4.Color wave effect.

You may use this filter very easily:

```WavesFilter waves = new WavesFilter(sourceImage);
waves.Period = 50;
pbxImage.Image = waves.ResultImage; ```

Selective Monochromatic Filter

Did you see the movie "Sin city"? This effect is similar to the visual design of the film.

Figure 5. “Sin City” effect.

It's very easy:

```SinCityFilter sin = new SinCityFilter(sourceImage, colorDialog.Color);
sin.Tolerance = 50;
pbxImage.Image = sin.ResultImage; ```

Statistic Color Correction

+   =

Figure 6. Statistic color correction.

We need two images - the image for correction and image-source of color. Example of code:

```StatisticColorCorrector scc = new StatisticColorCorrector(sourceImage, corrImage);
pbxImage.Image = scc.ResultImage; ```

Exchangeable image file format (Exif) is a specification for the image file format used by digital cameras (for all specifications, see here).
You may get a list of all EXIF tags:

```PropertyReader pr = new PropertyReader(bitmap);
foreach (EXIFTag tag in pr.GetAllTags())
{
// some code with  { tag.TagName, tag.TagValue }
}```

Figure 7. Exif tags.

Conclusion

This library can be useful for image processing, solution problems of skin detection, and creation filters. I use this library in my research very often.

Share

Software Developer
Russian Federation
Hello! My name is Maxim Subbotin.

Now I work in sphere of web-development. I'm interesting researches in SEO field.
If you interesting, you can see this tool:

KeywordCompetitor

 View All Threads First Prev Next
 My vote of 5 thatraja 1-Nov-10 1:57
 Last Visit: 31-Dec-99 19:00     Last Update: 25-Dec-14 12:29 Refresh 1