Research Teams

Research Teams

PEOPLE > Research Teams

Moongu Jeon

3-4-4 Project Director

  • Affiliation Gwangju Institute of Science and
  • Tel +82-62-715-2406
  • E-mail

  • Homepage

  • Research TopicPattern Recognition, Machine Learning, Computer Vision
Project : Development of Pow Power Image Sensor System Based on Color and Depth Information
Purpose ○ The purpose of our research is to develop an effective pow power image sensor system based on color and depth information for reducing power consumption level of conventional cameras.
○ Being combined with the conventional image sensor module such as surveillance camera and vehicle black-box, the low power event detection module wakes up the main sensor module(conventional image sensor module) when an event is detected.
○ The proposed system has the procedures for reducing power consumption. Only if the low power event detection module detects an event, it activates the main sensor module. Otherwise, the main sensor module is deactivated.
○ Also, we will implement more robust real-time applicable abnormal situation detection module using color and depth information from the main sensor module than using only color information.
Contents ○ For developing the proposed pow power intelligent image sensor system’s low power event detection module, 1) FPGA control chip, 2) low power image senor, and 3) Integration module should be implemented.Here is the research contents in detail.
- FPGA control chip : Implementing a developed intelligent image analysis algorithm and electric power(operating mode) control signal system on FPGA
- Low power image sensor : research and development of low power image sensor receiving the signal of electric power control chip as an input and sending low power image as output
- Integration module : Implementing the low power event detection module by integrating FPGA control chip with lower power image sensor.
○ The main sensor module has an abnormal situation recognition module based on color and depth information. the module is composed of 1) image preprocessing, 2) abnormal situation recognition. Here is the detail process as follows.
- Image preprocessing : Denoising and background subtraction of the image(color and depth) from the main sensor module to make abnormal situation recognition more accurate.
- Abnormal situation recognition : Research about spotting image sequence likely to have abnormal situation and classifying the spotted image sequence.
○ Finally, combining low power event detection module with the main sensor module, we will develop integrated Pow Power Image Sensor System Based on Color and Depth Information.
Expected Contribution ○ Generally, because the probability of meaningful event occurrence in video surveillance is much less than the probability of the opposite(normal) situation, the proposed image sensor system will allow us to save much power consumption.
○ Basically, this system can not only save maintenance cost of various images sensors but also make sustainable time of cameras using battery power much longer.
○ Being different with conventional color-based image sensor system, the main sensor module's color and depth image sensor can estimate distance between an object and the sensor. Thus, using both color and depth information, we will improve abnormal situation recognition performance much better than the case only using color.
○ The technology gap between Korean companies and the leading companies like IBM and AXIS in intelligent image analysis is growing every year. Moreover, the amount of export of Korean companies is decreasing. Under the circumstances, our proposed pow power image sensor system will help Korean corporations to be competitive by changing their main products from HW oriented CCTV/IP cameras to SW oriented image sensors.
No Title Year Phase
No Title Country Date Phase
No title Conference Name Date Phase
4 A New Performance Evaluation Software for Background Subtraction Algorithms ISCE 2014 2014-06-22 Phase 2
1st year
3 Development of an Emotional Gesture Recognition System in Dark environments ISCE 2014 2014-06-22 Phase 2
1st year
2 A Framework for Real Time Vehicle Pose Estimation based on synthetic method of obtaining 2D-to-3D Point Correspondence 한국정보처리학회 춘계학술발표대회 2014 2014-04-24 Phase 2
1st year
1 Codebook-GMM 모델을 이용한 배경제거 기법 IPIU 2014 2014-02-11 Phase 2
1st year

Warning: Unknown: Your script possibly relies on a session side-effect which existed until PHP 4.2.3. Please be advised that the session extension does not consider global variables as a source of data, unless register_globals is enabled. You can disable this functionality and this warning by setting session.bug_compat_42 or session.bug_compat_warn to off, respectively in Unknown on line 0