ASUS DITCamera project

Introduction

To detect a lens external condition, like dust, stain, illumination, ... ,etc, we build three algorithm to detect specific problem.

1. Shading

• Definition: The light fall-off observed towards the edges of an image.
• Reason:
  • Lens vignetting
  • The angular dependence of quantum efficiency due to use microlenses.
  • ... etc.
• Detect environment: White card.
• Detect method
  1. Evaluate highest pixel block ($\frac{1}{10}W\times\frac{1}{10}H$) average $Y_C$ from center block.
  2. Evaluate corner pixel block average $\text{Y}_1,\text{Y}_2,\text{Y}_3,\text{Y}_4$ ($\frac{1}{10}\text{W}\times\frac{1}{10}\text{H}$) .
  3. Evaluate shading (%) $\text{A}_i=\frac{\text{Y}_i}{\text{Y}_C}\times 100$, where $1\leq i\leq4$.
• Qualified condition
  • $\text{CenterLow}\leq\text{Y}_C\leq\text{CenterUp}$ .
  • $\text{PassLevel}\leq \text{A}_i \leq \text{PassLevelUp}$, where $1\leq i\leq4$ .
  • $\max_{1\leq i\leq4}{A_i}-\min_{1\leq i\leq4}{\text{A}_i}\leq \text{Diff}$.

2. Flare

• Definition: Unwanted light in an image that did not originate from the related scene object
• Reason:
  • Everything between the object and the camera's front lens, e.g., haze or dust in the air.
  • Interreflections between the optical parts of the lens and dust and dirt on the lens elements.
  • Light reflected from the edges and framing of the lens or the shutter, diaphragm, camera body, etc.
  • Light from the optical system itself with spherical aberration and chromatic aberrations etc.
  • Light leakage
• Detect environment: ANSI Contrast
• Detect method
  1. Statistics all pixel value distribution.
  2. Evaluate standard deviation $\sigma$ from pixel distribution.
• Qualified condition
  • $\text{StdThreshold}\leq\sigma$ .

3. Blemish

• Definition: Visible spots or marks in the image.
• Reason:
  • Sensor defects.
• Detect environment: White card.
• Detect method
  • Pass image to fast difference filtering
  • Pass image to filter to de-noise.
  • Binarize image by $\text{IntensityTolerance}$.
• Qualified condition
  • All pixel intensity $\text{Img}_{ij}<\text{IntensityTolerance}$.

Example

Download

git clone https://github.com/ed28311/DITCamera.git --recursive [Target Directory]

Build

cd [Target Directory]
mkdir build 
cmake ..
make 

Execute

Execute singal mission and output report.

./bin/main -{mode}{configMode} {config path} {image path}

Execute multiple mission and integrate to one report

./bin/main \ 
	 -{mode}{configMode} {config path} {image path} \
	 -{mode}{configMode} {config path} {image path} \
	 ... \
	 -{mode}{configMode} {config path} {image path}

Sample Batch file usage. (Windows)

launch.bat ${Base directory} ${File_path}

Structure

Tree

.
├── CameraSpec.spe               -> Basic config file.
├── CMakeLists.txt
├── launch.bat
├── launch.sh
├── README.md
├── LICENSE
├── str
│   ├── algorithmDispatcher.hpp  -> Assign task to correspond algorithm.
│   ├── configLoader.hpp         -> Load Config and constructure config class.
│   ├── DITAlgorithm.hpp         -> Basic algorithm class structure.
│   ├── DITAlgorithms.hpp        -> Header integrate all algorithm.
│   ├── DITConfig.hpp            -> Basic config class structure.
│   ├── logger.hpp               -> To log the result and save as file.
│   ├── method                   -> Directory to storage algorithm.
│   │   ├── blemish.hpp
│   │   ├── flare.hpp
│   │   └── shading.hpp
│   └── tool
│       └── displayer.hpp        -> To prettify string and print variable info.
└── test_pic                     -> Test picture.
    ├── blemish.jpg
    ├── blemishNG.jpg
    ├── blemishOK.jpg
    ├── flare.jpg
    ├── flareNG.jpg
    ├── flareOK.jpg
    ├── shading.jpg
    ├── shadingNG.jpg
    └── shadingOK.jpg

Class diagram