SysEng Computer Vison Systems
Download as PPT, PDF0 likes13 views
1. The document describes an automated vision system that uses image processing to inspect biometric passports and ensure logos and IDs do not exceed boundaries. 2. It also discusses using a vision system with cameras to monitor conditions at an outdoor site, such as detecting foreign objects, dealing with changing lighting, and transmitting images via GPRS for off-site analysis. 3. Another application discussed is using image processing to monitor levels of flotsam in a river in real-time, accounting for challenges like weather conditions, and determining rubbish amounts on site after calibration.
SysEng Computer Vison Systems
- 2. Automated Vision System in Controlled Lighting Conditions
- 6. Inspection of printing position of the Logo & ID for Biometric Passport that does not exceed the ‘Eclipse’
- 7. Images Captured Camera 1 ID Camera 2 Logo
- 8. Vision Inspection Method Boundary set for detection Detected ID & Logo shall not exceed the Boundary set
- 9. Automated Vision System in un- controlled outdoor lighting conditions
- 10. www.emonitoringsystem.com Last date is 30 Oct 2008 19:00 Hours User: lumchang Password: lumc Project was for Lum Chang KPE project PUB monitors this Contact Nor Hood
- 11. Gate Close Close Open Open Open Close Pump ON ON OFF OFF OFF ON Water 5 3 4 3 2 4 Level (@High=5,Low=1) System extract information about the Floodgate is open or close, water pump is working status and water level (complementing the hydrostatic water level sensors which are more accurate) The camera is a sensor with 640x480 pixels and 256 grey scale. This means that it has 307,200 8 bit sensors The images are stored for record purposes, zoom in and enhanced to extract more information, compensate for lighting so that they add value to any engineer analyzing the site conditions
- 12. PUB DTSS Illegal Activities Monitoring Introduction 1. Foreign object detection within restricted zone in an outdoor environment. 2. Lighting & shadow changing from time to time is the biggest challenge. 3. Advanced image algorithm will overcome the issue with the features of; • Images transmit through GPRS for backend analysis from site • Minimize false alarms caused by changing lightings & shadows. • Detect foreign objects under normal site conditions. • Highlight the foreign object on client screen on web. • Provide output & alert through SMS/Email foreign object/s been detected.
- 13. Methodology 1, Combined algorithms (Pattern matching 80%+ Color mean shifting 20%+ Strong white texture 10%) 2, Multiple dynamic reference frames enhance the result 3, Pixel size converted to object size Identify, range filtering 4, Preprocessing to reduce misalignment and color distortion Features of the POC software 1, Be able to detect weak signal 2, Not sensitive to shadow 3, Multiple dynamically reference images are used 4, Fast image alignment algorithm is incorporated 5, Range filtering is used to identify blobs
- 14. Test 1: Be able to detect object with weak signal and light variation
- 15. Test 2: Shadow did not affect detection result Shadow
- 16. Test 3: Highlight detected objects ROI Identified 2 objects Multiple references
- 17. SysEng (S) Pte Ltd By: Tan Chuan Chiat 16th Mar 2012 Progress Update on Flotsam Detection System
- 18. Test Site Installed at Kallang River on 9th Mar SysEng (S) Pte Ltd
- 19. Station Setup Solar Panel Camera Controller Box - Images are sent to server every 15 minute - Server process images and determine the rubbish amount SysEng (S) Pte Ltd
- 20. Image Snapshot EMPTY FULL FULL SysEng (S) Pte Ltd EMPTY
- 21. Image Processing Machine vision compute amount of rubbish in the flotsam in real-time and requires on site calibration SysEng (S) Pte Ltd 07 % 61 % 82 %
- 22. Challenges Weather condition: e.g. sunlight, rains Night time low light condition SysEng (S) Pte Ltd
- 23. Copyright of SysEng - Confidential 23 Data Logging system on an off-shore Floating Crane Measurements were carried out to determine the stresses and movements in space experienced by the hook. Strain gauges, X-Y-Z rotational sensors and X-Y Imaging Displacement sensors were used and the data were time synchronized. Contact-less techniques were used to prevent damages during field measurements. Image Processing Systems were used to track the hook movements in space, while the X-Y-Z rotational sensors were for rotation in space information. The strain gauges measured the stresses in the slings.
- 24. Copyright of SysEng - Confidential 24 Battery Powered High Speed Strain Gauge Data Loggers -16 simultaneous SG data acquisition @ 100 samples /sec/channel - Storage for 60 minutes data streaming logger Mains Powered Digital Image Recording Systems -5 frames per sec per logger - 1024 x 768 pixels per frame Battery Powered X-Z accelerometer s with Data Loggers -3 G with 10 mG resolution -200 samples/sec Synchronization 1 S Pulse FM Broadcaster to ensure accurate time stamping for all measured data recording Dynamic Crane Measurement System
- 25. Cars coming in Cars going out Number of free car park lots WiFi Camer a Number of cars in: 120 Number of cars out: 34 Number of Car Park Lots 240 Number of Free Car Park Lots 240 – 120 + 34 = 154 Percentage of utilization 154/240 = 64 % full When utilization is above 80% turn RED Real Time Image using OCR technology at server end in Real Time The same camera system is also used for recording for Security purposes for car park surveillance. There will be multiple WiFi and Cabled cameras around the locations for dual purposes
- 26. Using Imaging System for lowering costs in Open Car Park Space