JS-01

H.O. RETRO REFLECTOR

Existing RETRO REFLECTOR was a very expensive product because it was manually manufactured by depositing gold, silver, and aluminum on the glass surface.
However, we have the technology to do aluminum vacuum deposition by injection with ABS material.

RETRO REFLECTOR

  • A retroreflector is a reflector in which light from a light source is reflected off the surface of an object and returned to the light source, Retroreflection is characterized by reflecting light in the direction of the light source no matter what angle the light shines on the retroreflective material.
  • When a car's headlight or flashlight shines on a retroreflector
    It can be easily seen by people at the light source by turning it back in the direction the light came from.

light transmittance method

The light source emitted from the light source passes through the atmosphere on all optical paths between the light source and the detector,
At this time, the intensity of the initial light source changes due to the light absorption and light scattering materials (fog, fine dust, harmful gas) existing in the light path
It is detected through a photodetector, and the visible distance is calculated by applying the Lambert-Beer law to the intensity of the detected light source.

    JS-01

  • The light-receiving part and the light-emitting part are manufactured integrally, and the reflector is installed 50 to 100 m apart.

    JS-03

  • While improving the accuracy of the visibility distance measurement (40% increase compared to the existing accuracy) compared to the existing visibility measurement system,
    To develop a visible distance measuring system that simplifies installation (minimizes the required installation area)
  • *Acquired the patent for 'integrated multi-wavelength remote visibility measuring instrument' from Korea Research Institute of Standards and Science, and certified as a research institute in Daedeok Innopolis (Application No. 10-2016-0129455)
  • JS-01
  • JS-03

JS-01

  • The separation distance is twice as advantageous as using a reflector compared to the general light transmittance type product,
    Where the reflector is installed, there is no need to install electrical and communication cables.
  • LPOC(Long Path Open Cell)technology

  • light transmittance method

  • 530nm LED light source

JS-03

  • LPOC(Long Path Open Cell)technology

    The optical path is extended through the plurality of reflecting mirrors so that the optical path of the incident light is reflected and extended. As a result, the optical path of the incident light is extended and the spectral distribution of the incident light and the emitted light can be precisely detected as the frequency of contact with the atmosphere and gas is increased.
  • light transmittance method

    The light source emitted from the light source passes through the atmosphere on all optical paths between the light source and the detector, At this time, the intensity of the initial light source changes due to the light absorption and light scattering materials (fog, fine dust, harmful gas) existing in the light path. This is detected through a photodetector, and the visible distance is calculated by applying the Lambert-Beer law to the intensity of the detected light source.
  • 530nm LED light source

    The human eye is most sensitive to the green light spectrum. Therefore, this wavelength is particularly useful for pointer and measurement technology applications because it is brighter than other colors at the same power. It is easier and more effective to maintain than other wavelengths of color within visible light.

Development of weather measurement algorithm using labeled image data

01

Refine data set

  • Validate by constructing a 100% valid training set that is validated and augmented to correct the learning set and remove variations and labeling errors.
  • To induce excellent results in various environments for arbitrary data based on the learned data set

    This leads to excellent results in fault tolerance measurement and accuracy.

02

Redesign CNN training algorithm

  • Improving the flexibility and speed of learning queuing and data set balancing algorithms through image enhancement
  • Development of an algorithm that can process simple vector values ​​as input values ​​by simplifying the input complexity of the existing CNN learning algorithm

03

Perform hyper parameter tweaking

  • Optimization of model format to advance learning model and secure reliability (low memory consumption, fast inference)
  • Optimized by developing and optimizing parameter setting algorithms to minimize learning complexity and retain fast learning and reasoning capabilities

04

Compare results, Optimize format

  • Server for Native Tensorflow
    Derive the optimal format for artificial intelligence engine (Tensorflow) web service
  • By minimizing the complexity of multidimensional input values ​​in Tensorflow,
    Developed an optimization format of the input format to enable learning with only a single variable

05

Implement FT based version

  • Want to test implementation of frequency-based CNN model focused on regression analysis instead of classifier-centric CNN (generating continuous visibility values ​​instead of discrete classes)
  • Develop an algorithm so that learning occurs in the stage of optimizing the cost by applying the regression algorithm to the CNN

06

Implement DCP based verstion

  • Developed by applying a Fourier transformation-based image processing algorithm that can predict the amount of sunlight regardless of the contrast of data based on the darkness of the surrounding environment and the darkness between constituent pixels
TOP