CST engages in the research and development of computational sensor, machine vision, image tracking, image exploitation, and other novel technologies. Our R&D team consists of highly qualified researchers and experts. They have published textbooks and numerous refereed papers, edited special issues for technical journals, awarded patents, participated in organizing international conferences in their own research areas.
Our R&D work has been funded by the Department of Defense and other government agencies. The R&D projects include:
|
Study of a novel computing device, Neural Geometric Engine, for spatial vision including image registration and three dimensional recovery of motion and surface structure of scene |
|
Study of a computer vision system for landmark recognition for unmanned ground vehicles |
|
Integrated Signal Processor for Missile Terminal Seeker: Target Acquisition, Tracking, and Reacquisition |
|
Artificial Vision Geolocation System for determining 3-D geodestic coordinates of ground targets using aerial imagery |
|
USDA funded Machine Vision Application |
|
Ground Target Tracking via Rapid Change of Sensor Orientation |
|
Beyond RFID: Tagging Integrated with Image-based Object Tracking |
CST's first generation of Artificial Vision Geolocation System (AVGS 1.0) is a software package that takes aerial images as input and calculates the GPS coordinates of targets. The images are taken by commercially off-the-shelf digital cameras. AVGS 1.0 does not rely on any ground control point and determines the target geolocation at the accuracy of about one-meter.
The operator interacts with AVGS engine through a user-friendly GUI interface. Once the image of an interested area is loaded into the AVGS system, the operator can manipulate with the image to focus his intention into the smaller area around the target, as shown in the Area Selection Screen. One or more targets are then selected, as shown in the Target Selection Screen. A number of aerial images that cover the target area and taken at different camera stations are loaded into the system. The AVGS engine uses CST proprietary algorithms to register to feature points among these images and calculates the relative orientation of these digital cameras to the accuracy up to1 miliradian. Using the GPS coordinates and relative orientation information of at lease 3 cameras, the GPS coordinates of targets are accurately determined.
AVGS system has many applications in photogrammetry, computer vision, biomedical instrumentation, and smart weapon systems. AVGS 1.0 is expected to be released soon. If you need any information, please contact info@cstcorp.com
Area Selection Screen
Target Selection Screen
Automated Vision for Sex Separation of Poultry Chicks
The U.S. processes 160 million boilers each week. This project will develop an automated vision system to recognize the sex of poultry chicks and separate them into different bins which is currently done manually. The results from this project will advance the poultry operations with improved work environment, labor saving, feed conversion efficiency, and cost-effective production. The project is supported by USDA SBIR program.
Ground Target Tracking via Rapid Change of Sensor Orientation
 |
 |
|||||||||||||
|
|
|||||||||||||
 |
 |
|||||||||||||
|
|
|||||||||||||
|
||||||||||||||