This code repository attempts to implement a color model space and skin detection technique of two papers.
- A novel skin color model in YCbCr color space and its application to human face detection
- Skin Detection Based on Image Color Segmentation with Histogram and K-Means Clustering
I am planning to use this code to determine the skin color of patient with inflammatory skin conditions. By segmenting the diseased area and only considering the non-diseased area for skin color computation. The process follows these steps: training a novel skin color model, creating an initial mask to segment diseased skin, and creating a final mask. Using the first paper that I’ve mentioned, I created skin clusters in the YCbCr space. This let me compute the centroid and covariance of several clusters and a mean cluster. Once I developed the mean cluster that incorporated enough skin color, I began to run an initial inference to get a preliminary mask. Lastly, I used thresh holding techniques (Ostu's method) and experimental values found in the paper to further mask the image.
I initially used the implementation by kamiry, and later built upon it, which was an implementation of the first paper (1).
The following software requires that you have a copy of MATLAB R2020a (or greater) or a compatible version of Octave GNU. I used the following add-ons for MATLAB:
You need to specify the image directory in the main.m file. After which, you can run the main.m MATLAB script.
Figure 1. Images of orginal input image, first mask, final composite maks, and ITA calcuation on skin color patch respectively
 |
 |
|---|---|
| Figure 2.1. Image of gaussian ellipsoid plot for skin cluster | Figure 2.2. A 3D cluster plot of skin pixels of an image in YCbCr space |
Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.
Please make sure to update tests as appropriate.