January 31, 2007

Limiting Surveillance

From Wired News:

A short opinion piece on the long term implications/threats of surveillance. The last paragraph was very interesting to me:

We can choose to use highly invasive technologies only for the more serious security and law enforcement purposes. But we must build those restrictions into the machines themselves. Law alone will not be enough to control the natural human desire to use technology for all it is capable of doing, regardless of our values or goals. The inevitable alternative is mission creep.

I know vendors have sold smart camera systems that are capable of respecting various restrictions, e.g., such as the IBM system that allows images of individuals not under suspicion to be stripped out (link). But how many governments or corporations have asked/mandated these features when they purchase smart camera systems?

Posted by rshah at 10:02 AM | Comments (0) | TrackBack

January 27, 2007

Red Light Cameras Update

I haven't written much about red light cameras, so here is big post to cover a number of issues. Two recent articles that provide some nice background are from CBS and USA Today.

Redflex Traffic Systems is a leading provider of red light cameras. They have 40% of the market in the US. They have contracts with 90 cities in the US as well as photo speed programs in 8 states. The Chicago has 60 cameras from Redflex.

How the RedFlex System Works:

Redflex red light cameras will help cities nab about 4,000 red light runners per day or just under 1.5 million in 2006. Here's how they do it:
1. Road sensors: Road-embedded sensors detect a car moving toward an intersection. Sensors calculate the car's speed and determine whether it will run the red light.
2. Camera activated: Sensors send a signal to the overhead camera to begin recording images of the car as it nears the intersection.
3. 12 seconds of video shot: The camera records 12 seconds of video — 6 seconds before the car runs the red light and 6 seconds after.
4. Additional images captured: In addition to images of the car in the intersection, the system records a shot of the license plate and — in jurisdictions using driver liability and facial photography — an image of the driver.
5. Citation issued: Redflex checks images for quality and turns them over to the city's traffic law enforcement agency, which reviews it for a violation and mails a ticket to the vehicle owner.

Detecting Red Light Cameras. Cobra is developing new GPS radar detectors that alert drivers where traffic cameras are being used according to this. This appears to be a similar approach to the NAVTEQ camera alert system which maintains information on the location of cameras, see previous post.

Fighting Red Light Tickets in Chicago, from the Sun-Times:
Most people just pay the ticket. Only 10% of those fighting the ticket actually succeed. This article provides a bit more detail on fighting tickets and why people usually fail. As a result, by October of 2006 the city had already collected $12.5 million in fines for 2006.

Posted by rshah at 04:43 AM | Comments (0) | TrackBack

January 24, 2007

Latest on Gunshot Detection Technology

From SecurityInfoWatch.com:
A good article on the state of gunshot detection technology. It notes that 16 cities have installed ShotSpotter. A study in 1999 found the detection system to be accurate 80 percent of the time within 25 feet.

It also highlights some "success stories":

Less headline-grabbing are the cases seen in Minneapolis since installing ShotSpotter last month. Police have netted three felons, two semiautomatic guns, and recovered one stolen car. It also provided additional information in three shooting cases. "It's just a better compass. It still takes good cops, persistent investigation, and good police skills," says Lt. Gregory Reinhardt, spokesman for the Minneapolis police department. "It's just pointing us in a better direction."
However, Lt. Reinhardt admits that none of the arrested felons and confiscated items were necessarily involved in the original shooting. In one case, police arrived to find a car speeding off. Police pursued, then apprehended a suspect - a convicted felon - who tried to flee. In the car was a loaded semiautomatic pistol. In two other cases, police arrived to find people loitering. On each occasion they took names and found a person wanted on a warrant. "It's sort of hard to fathom that the purpose of the thing is to put police in a place where they can pick up people who are wanted on other warrants," says Mr. Yohnka.

A final interesting point concerns data security:

Data security will be one of the first questions. The entire system uses encryption, from sensor to server to dispatcher, says James Beldock, president of ShotSpotter. The server stores a record of each gunshot report that includes the time, the sensor readings, and calibration data.

Posted by rshah at 09:19 PM | Comments (0) | TrackBack

Privacy Issues with Smart Cameras

I thought I would highlight two interesting articles on privacy with smart cameras.

The first comes from Reason and is titled Is Privacy Overrated?
Its key insight is the gains that come from a transparent society ala Brin.

My credit card company has long known where I buy underwear, but I don't lay awake nights worried that prosecutors might demand knowledge of my preferences in skivvies. The ways in which that information can be accessed by the state are circumscribed by decades of legal precedent. We should remain vigilant that those precedents aren't eroded, and we should work to strengthen protections where necessary, but the collection of the information in itself is an unstoppable force, mostly for good--I like that I can sift thorough records ofeverything I have purchased in the last three years.

As this hints at, the article still notes that abuse of cameras by public officials is a reasonable fear and one that should be punished.

The second is by Bruce Schneier, the renowned security expert. He wrote an op ed for the Arizona Star. In the article, which discusses the use of ANPR, he points out how "technology is fundamentally changing the nature of surveillance." He uses the term "wholesale surveillance." He argues this switch is not a mater of degree, but fundamentally different than our previous notions of surveillance and privacy. He argues:

Wholesale surveillance is fast becoming the norm. Automatic toll-collection systems record when individual cars pass through toll booths. We can all be tracked by our cell phones. Our purchases are tracked by banks and credit-card companies, our telephone calls by phone companies, our Internet surfing habits by Web site operators.
The effects of wholesale surveillance on privacy and civil liberties are profound; but, unfortunately, the debate often gets mischaracterized as a question about how much privacy we need to give up in order to be secure. This is wrong. It's obvious that we are all safer when the police can use all techniques at their disposal. What we need are corresponding mechanisms to prevent abuse and that don't place an unreasonable burden on the innocent.

As a practical example, he suggests that ANPR systems should erase the data on innocent car owners and not save it. Furthermore, he argues that with automatic detection, think red-light cameras, we need to realign the detection and enforcement actions. He suggests removing criminal penalties for both red-light cameras and speed-trap cameras. Instead, the the cameras should issue citations without any "points" assessed against the driver.

Posted by rshah at 01:44 PM | Comments (0) | TrackBack

January 18, 2007

Google for Cameras

From Chicago Tribune:

A story describing 3VR's search technology for camera footage. I noted this technology about a year ago when it received quite a bit of press. Here is a bit more detail on how it works:

We discard everything except the best two or three frames," he said. The system looks for images that are similar to each other, Ross said, and this produces many results that aren't right on target. "But if you get 100 images to look at and 10 of them are what you want, that's good enough," he said.

A system costs $4,000 to $16,000, depending on its analytical ability, according to 3VR. The company, which has financial backing from an arm of the Central Intelligence Agency, collects information associated with video images, Ross said.

It notes if there is motion, if a human appears to be moving and which way. It notices if something in a scene changes and when. All of this information can be reduced to bytes in a database that's quickly searchable. That serves as a sort of index to the video images associated with each observation. And the system becomes better with practice.

After being told that 10 different images are all of the same hotel employee, Strand said his system can locate that employee viewed on any of the Talbott's cameras with about 80 percent accuracy.

Posted by rshah at 02:55 PM | Comments (0) | TrackBack

January 06, 2007

Smart Cameras at John Hopkins

From the Baltimoresun.com:
Some useful information on the smart camera system at John Hopkins University. Back in May 2005, I noted this camera system. The system relies on 89 cameras. The software is by Cernium and displays 19 scenes on a large screen in the command center. It is also sends alerts from 18 different behaviors including "people moving very fast or loitering; cars that stop suddenly or drive too fast; crowds that gather or disperse; unattended objects and people who fall". The article provides a number of examples of how these alerts have proved useful, such as detecting loitering:

A lone man is looking up and down the street, apparently waiting for someone. A pickup truck drives up. The man says something to the driver, gets in and they drive off. Minutes later, a block away, a woman is robbed at gunpoint by two men who speed off in a pickup. No one at the scene can describe the truck to campus security officers or to Baltimore police.

Hopkins' security system caught the robbery suspects on a video camera on Lovegrove Street. The software registered the man's behavior and the late hour, and alerted the security officer on duty in the command center. The view down Lovegrove was singled out by the computer amid the incoming imagery from 89 campus security cameras. It popped automatically onto the officer's screen, with the man's image highlighted in a yellow box. She quickly zoomed in and recorded images of the suspect, the truck and its license plate.

After the victim reported the robbery, the tag number led police within hours to a borrowed truck and the suspect, who had a police record. The victim picked him out of a photo lineup. He was arrested a few days later, linked to a second crime and charged with both.

"If we didn't have this video system, or she didn't focus on him, he would have gotten away," said Edmund Skrodzki, executive director for security for Hopkins' Homewood campus. "One person can't monitor 89 screens. You need help with it, and behavioral recognition provides that assistance."

• In addition to the Lovegrove loiterer, the system has alerted security to a juvenile as he attempted to steal a motorbike, leading to an arrest.
• Another youth was spotted, tracked by cameras and arrested after spray-painting graffiti on campus buildings.
• A nighttime bicycle thief was confronted after another alert brought nearby security officers to the scene. The crook dropped his bolt cutters and ran off, but the bike was recovered, Skrodzki said.
• Other potential criminals were warned off because of automated alerts as they cased a sorority house, or tried doorknobs and car handles near campus.
• Campus bike thefts dropped from 25 during the 2005 fall semester to three last fall. Overall crime was down 20 percent in 2006. And Skrodzki credits the behavioral recognition software for providing a critical assist.

Posted by rshah at 08:50 PM | Comments (0) | TrackBack