Machine Vision/Image Processing offers unparallel benefits over traditional sensors and measurement tools because it offers unique abilities not found in many traditional tools. Consider, for example, a conveyor with items traveling at 300 parts per minute, and you want to take a simple width measurement of the items coming down that production line. With machine vision, one can inspect every single part instead of randomly sampling and just hoping that the rest of the parts are OK. Because machine vision tools are nondestructive, you ensure that none of the parts are damaged in the measurement process while achieving extremely high-accuracy measurements.
We have a strong expertise in the area of machine vision/image processing. With the embedded systems and image processing as core technologies, we offer a total solution service covering R&D and product development. We have designed and implemented sophisticated systems involving the use of various imaging concepts and algorithms.
Snapshot of our expertise,
- Platforms - TI DSPs - DM642, DM6446, DM6437
- Image Enhancement - Adaptive Image Filtering, Histogram Equalization, Contrast Stretching, Dynamic Range Manipulation, Feature(s) based selective image enhancement,
- Image Segmetation - Thresholding (Global, Local), Region Growing, Watershed, Edge based Segmentation, Fuzzy Logic based edge detection/segmentation
- Tools & Utilities - Code Composer Studio v3.3 IDE, Visual Studio, .Net, Ultra-Edit, VEW, VSS
- Industrial Automation devices: PLC, HMI, CNC, DCS, SoftPLCs, M2M Systems
- Industrial Protocols: Modbus, Profibus, Foundation FieldBus, BacNET, CANbus, DeviceNET, HART, AB DF1
Inspection of Electronic Chips embedded in a moving Tape
- The aim was to inspect electronic chips embedded in a moving tape in real time – complete software development.
- Complete system in C language.
- 4 Chips embedded in a row and multiple rows on tape
- Tape is moving from one spool to another
- 3 Cameras installed above moving tape
- Merging of images as two camera capture 1.5 chips image
- Compare input images with master image based on client rules
- Declare in less than 2 seconds which Images / chips are OK
- Not OK chips are punctured by attached Robot arm
- Generate the pattern and statistics of faults for process improvement
- High reliability needed
Project successfully accepted by the client having 97% fault recognition with 100% reliability.
|
License Plate Recognition
- The aim was to develop a vision-based License Plate Recognition system that detect and read vehicle license plates for vehicle access control, parking lot billing, and traffic lane and security applications.
- Detection and extraction of the license plate is performed using innovative and robust image processing techniques that enable the system to read plate numbers even if lighting conditions are non-ideal or when characters are partially occluded by dirt or shadows.
- The Optical Character Recognition (OCR) module is based on a proprietary neural technology.
Features:
- High recognition rate
- High recognition speed
- Video or AVI file is used as recognition input source
- Most types of plates it can handle
- Widest range of picture quality it can handle
|
|
