July 7, 2005 1:52 PM PDT

Tech's part in preventing attacks

How do you pick out the terrorist in the crowd? That's the problem, underlined by the London explosions, that Pixlogic is trying to solve.

The Los Altos, Calif.-based company has created software that uses visual pattern recognition and search technologies to match archived still or video images with pictures gathered from security cameras or other sources. The software can also pick out anomalies--someone walking with a large box, or a truck that keeps coming back to the same spot--in hours of video footage. These features could all speed up intelligence work.

"It does a reasonable job of matching people that sort of look alike," Pixlogic CEO Joseph Santucci said. "Most (competing) software tools only work under constrained circumstances."


Pixim's technology aims
to avoid overexposure.

While tools for security and intelligence gathering have been a major focus for the technology industry since the Sept. 11 attacks, the bombings in London on Thursday will likely add further impetus to find and deploy such products. Although no technology can promise foolproof security, progress is being made on tightening up protection.

Various intelligence agencies have licensed Pixlogic's technology, for example, and In-Q-Tel, the CIA's venture arm, is an investor in the developer.

"We have put a little over 100 technologies into the intelligence community that are actively being used," Gilman Louie, CEO of In-Q-Tel, said in a recent interview. "Some of them you know, while other technologies we don't broadly advertise."

The different technologies that could help prevent events like those in London, or help apprehend terrorism suspects, can be grouped into a few basic categories:

Translation
Intelligence agencies around the world continue to face a shortage of speakers of Arabic and other languages often associated with terrorist groups. Translation can also be time-consuming.

Related photos
The world witnesses
Citizens share photos of the London blasts on sites around the Web.

To this end, Language Weaver has developed machine translation tools that can dynamically translate Arabic, Chinese and a few other languages into English. In its sales presentations, the company has its software produce an English manuscript of an Al-Jazeera broadcast while the broadcast is airing.

The company said it will release a Farsi version later this year.

"It used to be finding a needle in a haystack. Now it's trying to find a needle in a haystack in a field of haystacks," said Language Weaver CEO Bryce Benjamin. "There is a lot of focus on getting automated tools."

Search and text mining
Several companies are trying to devise tools that will find relationships between different pieces of information in a sea of seemingly unrelated, and largely unstructured, documents--a problem that's often compounded by translation difficulties.

CallMiner's analytics tool promises to convert recorded phone calls into text and then help agents try to establish relationships between the recorded conversation and data in existing intelligence flies. The system will pull out matching names as well as help analyze trends in large volumes of calls. Like most companies in this area, CallMiner counts both intelligence agencies and ordinary commercial ventures as customers.

In addition, Basis Technology has created a search engine that works on native Asian and Arabic languages, which cuts down on errors introduced through translation.

Other companies, such as IBM and MetaCarta have developed systems to search on geographic details in documents or e-mails. Ideally, this could let agents discern a person's movements.

Visual Identification
Like Pixlogic, A4Vision is working on 3D facial-recognition systems that could be used in a broad variety of security applications.

While Pixlogic and A4Vision have developed software for camera systems, Pixim has developed an image processor that cancels out noise to improve image quality. The company won a big contract with a bank after a camera installed in one branch captured a clear image of a bank robber that ordinarily would have been washed out by glare.

Commercial clients are also interested in this technology. Panasonic, for instance, uses a simplified version of Pixlogic's libraries in software that searches for images in family photos when consumers submit a text query ("Find Mom").

Monitoring
Both Dust Networks and Crossbow Technologies have worked for the last several years to tune systems that can digest information coming from several sensors or cameras. Dust recently created a "smart building" in Berkeley in which the network can actively monitor and control energy consumption. Both are also working on military projects.

Bio assay
Earlier this year, high-school student David Bauer won the Intel Science Talent Search for devising a way of exploiting quantum dots, which are florescent crystals, to identify an enzyme found in neurotoxins.

Some believe that nanotechnology will play a pivotal role in developing chemical sensors. In nanotechnology, a few atoms can do a job that now might take hundreds of thousands of atoms. That means that biohazard sensors using the technology could become far cheaper, smaller and easier to deploy.

Existing companies working in this area include IatroQuest and Seahawk BioSystems.

 

Join the conversation

Add your comment

The posting of advertisements, profanity, or personal attacks is prohibited. Click here to review our Terms of Use.

What's Hot

Discussions

Shared

RSS Feeds

Add headlines from CNET News to your homepage or feedreader.