AI Note: computer vision and color model of the lighting model

Lighting model

Flux

• Able to see around the world has two components: a source outside the case, the surface of the light source
• It refers to the light flux perceived by the human eye can radiated power, which is equal to the product of the radiation of a wavelength band per unit time and the view of the relative band
• In notation, LM units (lumens)
• 1lm = 0.00146 watts

Typical values ​​of different light sources

light source	Flux	Explanation
sun	3.566 * 10 ^ 28 lm	Energy source of our existence
Candlelight	12.56lm (4π)	Definition of flux
Incandescent / tungsten halogen lamp	12 ~ 24lm / W	More efficient than incandescent tungsten halogen lamps
Gas discharge lamps and fluorescent lamps	50~120lm / W	Gas discharge lamp
LED lights	110lm / W	no

radioactivity

• Irradiance: means to project radiant flux density (we can feel the intensity of light) on a planar surface. Refers to a surface plane to reach the unit time, the radiant energy per unit area
• The symbol E represents, typically in units of Lux (lux)
• 1lux = 1lm / m^2

Some typical irradiance

Scenes	Illuminance (lux)	Explanation
Night	0.001 ~ 0。02	0.02 to 0.3 moonlight
Indoor cloudy	5 ~ 50	Outdoor cloudy
Sunny indoor	100 ~ 1000	In doing visual system needs to fill light
Sunny sunlight	100000	You do not need to fill light in the visual system
Suitable for reading	300 ~ 750	Need 50 to 60 read books
Home video camera standard illumination	1400	In doing need to fill light vision systems

Common light source

According to different classification

• Direction: direct light, scattered light (diffusing plate)
• Spectrum: visible light into visible light
• Other: polarization, other

About backlight

• From the back light source illuminates the object
• Features: light emitting surface is a scattering surface, good uniformity. Specularly reflective material may be used, such as a flaw detection on a wafer or a glass substrate; detecting the LED, smiling electronic element size, shape, target test

Color light source

• Similar to the color filter (white or weakened)
• Strengthen the complementary color (black or strengthen)
• Features: If you want to highlight some of the more clearly the color, select the color corresponding to the complementary color of the ring light source for illumination, which can significantly improve the contrast of the image
• Examples: PCB board green background, with white light, MARK point contrast is not high enough, and the background mixed together easy, difficult to distinguish. With red light green background to improve the contrast, MARK point clearly visible.

Color Model

• RGB model
• CMYK model
• HSI model

Related to achieve corresponding color OpenCV

1) color space transducer means

• c ++ realize

  // 第一个参数：输入的图像
  // 第二个参数：转换以后的图像
  // 第三个参数：code: COLOR_BGR2GRAY, COLOR_BGR2HSV 表示在哪两个空间进行转换
  void cvtColor(InputArray src, OutputArray dst, int code, int dstCn = 0);
  

• python achieve

  dst = cv.cvtColor(src, code[, dst[,dstCn]])
  

2) channel separation

• c ++ realize

  // 第一个参数：原始多通道的图像
  // 第二个参数：通道分离后的结果，通常是个数组
  void split(const Mat& src, Mat *mvBegin);
  

• python achieved: the form: bgr [:,:, 0] represents the 0th channel (Blue) extract

3) sample program

Version 3.1 Cpp

#include "opencv2/opencv.hpp"
using namespace cv;

int main(void) {
    char *fn = "D:\\lena.jpg";
    Mat image1 = imread(fn);

    Mat image, gray, hsv, hsvChannels[3];
    pyrDown(image1, image);
    cvtColor(image, gray, COLOR_BGR2GRAY); // 把原始图像转化成灰度图像

    imshow("source image", image);
    imshow("gray", gray);
    waitKey(); // 不加这句话，窗口会闪退

    cvtColor(image, hsv, COLOR_BGR2HSV); // 把原始图像转化成hsv图像
    split(hsv, hsvChannels); // 通过split函数把hsv三个通道分离开来
    imshow("Hue", hsvChannels[0]); // 色度通道
    imshow("Saturation", hsvChannels[1]); // 饱和度通道
    imshow("Value", hsvChannels[2]); // 亮度通道
    waitKey(); 

    return 0;
}

3.2 Python version

import cv2 as cv

filename = r'/Users/johnny/Downloads/1.jpg'
img = cv.imread(filename)
gray = cv.cvtColor(img, cv.COLOR_BGR2GRAY)

cv.imshow("source image", img)
cv.imshow("gray", gray)
cv.waitKey()

hsv = cv.cvtColor(img, cv.COLOR_BGR2HSV) # 将bgr转化为hsv

# 显示hsv的各个通道
cv.imshow("Hue", hsv[:, :, 0]) # 在最后一维指定通道
cv.imshow("Saturation", hsv[:, :, 1])
cv.imshow("Value", hsv[:, :, 2])
cv.waitKey();

# 显示bgr的各个通道
cv.imshow("Blue", img[:, :, 0])
cv.imshow("Green", img[:, :, 1])
cv.imshow("Red", img[:, :, 2])

cv.waitKey()
cv.destroyAllWindows()