Objectives: To perform a more accurate analysis using exact rgb values

SkinTone 2.0

Objectives:

• 
  To perform a more accurate analysis using exact RGB values.
  
• 
  Take the control of the coordinates matching in algorithm side.
  
• 
  Automatic chart analysis using the white divisions from the chart.
  
• 
  Feed algorithm other charts and provide good accuracy with minimum/null algorithm changes.
  

Patterns detection in chart image

From a color chart with different tones and divided by white lines, algorithm automatically detect the areas and make an RGB coordinates average over the detected areas.

RGB matching

Then input skin images are analyzed and match with the chart analyzed image.

• 
  The input skin image analysis is performed using the same skin surface detection developed and used at Wrinkles 2.0, rescue map, and areas detected as external surface is taking for RGB color average (Please refer to Wrinkles 2.0 algorithm report for detailed information CH-RD-0001-19-02-19-Skin_Wrinkle_Algorithm_V_2.pptx).
  

Then the closest RGB points are matched using Euclidean distance over the RGB coordinates obtained from the input skin image compared with each of the detected regions in the chart as used in Fitzpatrick 2.0 algorithm (please refer to the Fitzpatrick 2.0 report for detailed information CH-RD-0002-19-03-29-Fitzpatrick_V2.0.pptx).

Testing Images

In the green boxes there is the RGB average obtained from the input skin image

Control (solid) images

(you can check the RGB values with the matching color tool in input image and chart surrounding the yellow mark)

Real skin images

Other samples of 2 analyzed images

Parameters 1.X.X (Local)

1. 
   For first image analysis (case if there is only one image depending on scenario) use the default parameters.
   

2. 
   If there are more than 1 analysis (like forehead and cheek):
   

• 
  Keep the previous output image (marked) the algorithm is validated for this case and input as chart image.
  
• 
  Change the point color.
  

3. 
   With more analysis keep the same logic re-using the previous output generated file.
   

Parameters 2.X.X (Cloud) -> API

Output

String containing the following values:

R, G, B, L, A, B, H, S, V