Imaging in Government Security. . . .by Nello Zuech, Contributing Editor

Had the opportunity to attend the Government Security Expo at the Washington Convention Center in Washington, DC held the week of April 24. While there not many exhibitors were there demonstrating intelligent video techniques of one type or another, there were a few. The show itself included a hodge-podge of exhibitors – everything from weapons, night vision devices, uniforms, police departments recruiting to bomb handling robots to emergency vehicles of one type or another.

What probably made attending the show even more interesting was Dr. Bob Shillman’s announcement during the Cognex quarterly analyst meeting that they are in the final stages of acquiring a company with products based on the same infrastructure technology found in machine vision but targeted at a non-industrial vision market. Given the suggestion was also made that the new product line would complement their existing person counter/tail gating product suggests that Cognex plans to enter the security market with a broader line of intelligent video products.

In any event, in the case of intelligent video at the Show it is apparent that more intelligence is being embodied. Perhaps the most impressive was technology from Australian-based iOmniscient. One application of their IQ-Infinity product they promoted was counting people in a crowd based on a camera mounted overhead. The suggestion was they could even give a reasonably accurate count when people were densely gathered. This product lends itself to assessing the length of a line of people and determining time to transition through a given point. They also offer a version capable of counting cars and also the ability to detect people or vehicles traveling in the wrong direction.

Another one of their interesting capabilities was what they called non-motion detection. This is targeted at detecting suspicious packages that remain in a fixed location for an extended period of time even in a very busy environment and despite constant motion in front of and around the object. The suggestion is that the system can detect an abandoned object that covers 2 x 2 pixels in the field-of-view of the camera. This system lends itself to detecting vehicles that are stopped because of breakdown or accident or detecting cars that are parked in no parking areas. By flipping the logic, they can monitor art museums to detect changes that might reflect the absence of a painting or another work of art from where it should be.

Another capability offered is behavior analysis – the ability to detect abnormal behavior such as running, loitering or someone slipping or falling. Similarly it could detect a suspicious car – one traveling slower or faster than normal traffic. The software can also be customized to detect specific patterns such as a U-turn or a person walking from car-to-car in a parking lot.

Some of the features of their system include:

• Ability to cope with environmental factors and dynamic lighting
• See perspective in a single camera
• Operates on a wireless PDA
• Copes with camera shake
• Works with existing camera infrastructure – analog or IP cameras

UK-based Vigilant Technology was another company offering products with intelligent video analysis. Their digital video recorders and integrated video servers incorporate behavior analysis and advanced search capabilities. The video analysis detects, classifies and tracks objects based on specific behaviors. Using the internal digital video motion and object tracking features, their system allows users to effectively retrieve and analyze information according to their criteria.

Vigilant’s activity detection features include:

• Continuous, scheduled, or event-triggered video and audio recordings from unlimited cameras in real time, with flexible frame rates
• Compression techniques enabling vast amounts of data to be stored online, long-term, in minimal space - reducing cost of ownership
• Instant identification of critical events…easily played back, analyzed, and sent via internet and intranet to any location
• 100% detection coverage -
  • VMD (Video Motion Detection) - activates an alarm whenever any change in a predefined designated area is detected
  • Audio Activity Detection - activates an alarm whenever there is any change above the background audio
  • Video Loss Detection - activates an alarm whenever there is loss of video, due to camera malfunction, cut wire, etc.
  • Camera Tamper Detection - activates an alarm whenever any type of tampering is detected, including blocking, re-positioning, or removal of cameras

PureTech Systems was another company with an interesting spin on intelligent object detection and tracking. After an object is detected, the software is able to track the object automatically with a Pan-Tilt-Zoom camera. The algorithms adapt to changes in lighting and irrelevant motion such as the movement of trees, etc. The system can be configured to detect objects of a user-specified size, speed and direction, while ignoring objects that do not meet the criteria specified.

The interesting spin they offer is the ability to integrate the event detection with a geographic information system, for example, automatically relating the source of the event to the geographic location of the event on a map and to the live video of the event. The object can then be integrated with the display of the street and area immediately surrounding the object being tracked. This geographic map-based command and control user interface makes it possible to dispatch personnel to specific known streets or intersections.

Safeguards’ Video TotalTrack® includes two modules, IntrusionTrack™ for intelligent intrusion detection by fixed cameras, and PTZTrack™, an innovative solution transforming manually controlled Pan/Tilt/Zoom cameras to automatic intruder tracking systems. This enables a user to lock on to an intruder and automatically keep the intruder in center focus regardless of its movements over the entire range covered by the camera. TotalTrack® can be connected to any type of camera: color, black & white, thermal or infrared.

Market Intelligence News & Insights: 

How Will Apple’s $500 Billion Investment for US Manufacturing, Education, and AI Play Out?

On February 24^th, Apple announced a “commitment… to spend and invest more than $500 billion in the U.S. over the next four years”, with plans for new manufacturing plants, enhanced support for manufacturing partners via their Advanced Manufacturing Fund, and a myriad of other initiatives. As part of this plan, Apple projects 20,000 new hires, with roles focused on R&D, silicon engineering, software development, and AI/ML engineering.

Read More

While there were a number of access control companies exhibiting including those with various door styles, none were promoting tail gating systems of any type. Most suggested that their systems detected tail gating by not permitting anyone who has not legally entered the area to leave the area without reconciling the detection of an illegal entry. This approach leaves a lot to be desired in this day and age, given suicide bomber mentality where they have no intention of leaving once they have entered the premises.

The show did have a number of companies exhibiting biometric recognition systems: Identix (finger, handprint and facial, DynCorp (finger vein pattern) and LG Electronics (iris). Northrup Grumman was there with their machine vision-based license plate reading technology that originated with Perceptics.

The other major imaging-based technologies exhibited related to x-ray-based systems of one type or another. American Science and Engineering employs what they call Z Backscatter techniques, which captures data from X-ray photons that are scattered from the object undergoing inspection. This primary scattering effect is known as "Compton Scattering." X-ray photons scatter differently when they encounter different types of materials. Compton scattering is material-dependent, with the lower atomic number materials scattering more strongly than the higher numbered ones. (Higher atomic number elements are more likely to absorb X-rays, either before or after being scattered.) Organic matter, specifically explosive material, contains low atomic number (low Z) elements such as carbon, oxygen, hydrogen, and nitrogen.

Their Flying Spot technology allows the position of the X-ray beam to be defined at every instant of time, so that any Z Backscatter signal received is easily correlated with the particular region of the cargo undergoing inspection. In addition to offering systems for baggage, parcel, letter and people inspection, they offer systems that can inspect cars, vans, and trucks, as well as palletized cargo, and air and sea cargo containers for:

• Illegal Drugs
• Illegal Immigrants
• Plastic Weapons and Explosives, including car and truck bombs
• Radioactive Threats, including nuclear devices and dirty bombs
• Smuggled goods, such as alcohol, tobacco products, and other legal goods smuggled to evade duties (trade fraud)
• Weapons or other inorganic threats, including metal weapons and shielding to conceal radioactive materials

While the ultimate decision is made by a person viewing a display, these systems do quite a bit of signal processing on the sensor inputs as well perform image enhancements to provide the sharpest images of anomalous conditions that could reflect concerns.

Rapiscan (an OSI company), in addition to offering systems for baggage and parcel inspection, also offer systems for cargo and container inspection. Their systems use linear accelerators to penetrate even the densest cargo. Versions are truck mounted or gantry mounted. Systems are able to inspect a 20’ container in 30 seconds. They also offer systems for inspecting baggage and cargo that goes into the hold of plane or ship. System uses five distinct views of the piece each based on dual energy x-ray beams. Suggest they can inspect up to 1800 bags per hour. The remote operator analysis is aided by the identification of the potential threat items by automated threat detection algorithms. By analyzing the five views the systems detect presence and position of a threat based on context, material density, size and effective atomic number. It was observed that, while this system is being deployed in Europe and elsewhere, it is not being used in the US because DOT has dictated that only CT-based systems can be used for these applications.

L-3 Communications and GE/Invision are two companies currently offering CT-based systems for inspecting bags that are placed in the holds of airplanes. As the conveyor moves each bag through the machine, the system creates a scan projection X-ray image. The software analyzes these images and compares their CT properties with those of known explosives. If a match is found, the system alarms and displays the object on the screen for an operator to review.

L-3 Communications’ Security and Detection Systems group also offer conventional x-ray inspection systems for baggage and parcels as well as container inspection systems able to inspect up to 25 trucks per hour. Systems offer pseudo-color overlay, variable contrast control, edge enhancement, gray-level mapping, log and square root mapping, histogram equalization and their “DeepScan imaging algorithm.

Astrophysics, a name from the past, also exhibited their parcel and baggage inspection systems. Principals came from old EG&G Astrophysics, a company with a long history in this business. At one time this group offered machine vision-based x-ray systems for the food and packaging market. However, the resurrected company does not offer these systems. Systems generally include automatic edge segmentation, geometric distortion correction, organic/inorganic imaging and ability to measure the atomic weight within a user specified region of interest. One of their versions is able to inspect an entire pallet of parcels or bags at one time.

Other companies offering, more or less, traditional x-ray-based systems for the parcel and baggage handling market included: AutoClear, and Smiths Detection. Most offer image enhancement capabilities and ability to inject “phantoms” of potential threats into the image to keep the inspectors on their toes.

All-in-all, in spite of the fact that not as many imaging-based companies exhibited that probably should have, it was a very interesting display of how machine vision-like capabilities are migrating into the security and surveillance market. Given the more widespread demand for security than for automated inspection the market for computer vision-based technology in the security and surveillance market will soon exceed that of the machine vision market in North America.