Color Image

cv

我们只需要三个参数就能描述任何一个色彩并非基于任何物理上的原因，而是因为人类眼睛有对于短（S）、中（M）和长（L）波长光的感受器（视锥细胞），所以原则上只要三个参数便能描述颜色感觉了。一些人拥有四种感受器，因此他们比普通人能感觉到的颜色多得多。

这个色彩空间可以由多种不同的三原色来表示，且根据格拉斯曼定律（Grassmann’s Law），人类对色彩的感受是线性的。这意味着色彩空间又是一个向量空间，而每种三原色都作为一组基来张成这个色彩空间。

Color imaging has been implemented in a number of ways:

• Field sequential (场时序).
• Multi-chip (多芯片).
• Color filter array (色彩滤镜阵列).
• Foveon X3 sensor

1 Subtractive color mixing

2 Color matching

Why specify color numerically?

• Accurate color reproduction is commercially valuable—Many products are identified by color
• Recognize color names
  
  • 11:black, blue, brown,grey,green,orange,pink,purple,red, white, and yellow in English.
  • Color reproduction problems increased by prevalence of digital imaging(数字成像的普遍存在增加了色彩再现问题)
    
    • How to ensure that everyone perceives the same color?
    • What spectral radiances produce the same response from people under simple viewing conditions?

3 Color Space

3.1 Linear color spaces

• RGB
  Single wavelength primaries. Good for devices (e.g. phosphors for monitor), but not for perception.
  

• CIE XYZ(国际照明委员会)
  $(x,y)=(X/(X+Y+Z), Y/(X+Y+Z))$
  X,Y,Z称为三色刺激值,实现起来很简单：对于一个颜色C=(X,Y,Z)，三个参数同时除以X+Y+Z进行归一化，设改变后的坐标为(x,y,z)，由于z=1-x-y，只需用(x,y)即可表示这个颜色的色度，接着我们用Y=Y/y代表明度（注意Y和斜体Y含义不同），这样C=(X,Y,Z)=(x,y,Y)。几何上看，相当于是将向量(X,Y,Z)投影到它与平面X+Y+Z=1的交点(x,y,z)。将所有的颜色都转化到了这个面上，由于x,y,z>0，实际还在一个三角形内部。最后，再将这个面投影到xoy坐标面，即可得到下面这张图，称为色度图，其外侧是单一波长的光（单色光）轨迹。
  
  CIE XYZ is not a uniform color space, so magnitude in coordinates are poor indicator of color “distance”.
  

3.2 Non-linear color space

• HSV
  Hue, Saturation, Value (Brightness)色调，饱和度，值（亮度）
  
• CIE LAB
  
  

There are many color spaces for different using, such as, RGB, CMY, XYZ, xyY, I1I2I3, UVW, LSLM, Lab, Luv, LHC, LHS, HSV, HSV Polar, HSI, HSI Polar, YUV, YIQ, YCbCr.
These color spaces can convert between with each other.

YIQ system is the color primary system adopted by National Television System Committee (NTSC) for color TV broadcasting, (PAL).

4 Color as a low-level cue for CBIR

5 Color–based image retrieval

Algorithm Step:

• Given a collection (database) of images:
  
  • Extract and store one color histogram per image
• Given new query image:
  
  • Extract its color histogram
  • For each database image:
    
    • Compute intersection between query histogram and database histogram
  • Sort intersection value (highest score=most similar)
  • Rank database items relative to query based on this sorted order

6 Color-based skin dtection