Criminal? We'll let the video decide

The New York Times has an article by David Halbfinger about the use of video in criminal prosecution:

So much for the camera never lying. The arrests of protesters who seized a building at the New School on Friday turned into something of a race to YouTube, as sharply contrasting videos recorded by the police and civilians conveyed vastly different impressions of what was going on— and, more specifically, just who was doing what to whom.
As the twenty-two people who were arrested were arraigned in Manhattan Criminal Court on Saturday afternoon (all were released on their own recognizance), law enforcement officials and civil liberties advocates alike sought to take stock of the dueling videos. They are the latest instant artifacts of a society in which cameras are seemingly omnipresent, but the truth can still remain elusive.
A police videotape, presumably made by the department’s Technical Assistance Response Unit and posted at its site on YouTube, shows officers calmly entering a room where protesters are seated; the protesters helpfully remove their backpacks and extend their arms to make it easier for the police to handcuff them. The level of politeness on either side seems more appropriate to a tea party.
But two videotapes recorded by people on Fourteenth Street captured a vastly different mood. One, the work of Brandon Jourdan, a freelance journalist, shows officers forcing a door shut against resistance, and spraying pepper spray at those on the inside. It also shows an officer pushing a shouting man in the face and knocking him to the ground before several officers arrest the man— seemingly without provocation.
But yet another videotape, identified on the New School Free Press Web site as the work of a videographer named Chris McCallion, shows demonstrators scuffling with police and at least one of them heaving a metal barricade toward the officers. (The Police Department released a lengthier video from a similar angle, also showing a glimpse of the barricade and a darkly dressed hurler, who scampers away as officers race after him.) Taken together, it seems the throwing of the barricade may have prompted officers to chase a group of protesters down the street, including whoever threw it, and perhaps including the man who was knocked to the ground.
None of the videos, however, tell the whole story. “The myth about film is that it’s objective, that film is the camera’s eye, so it’s not prejudiced,” said Paul Chevigny, a New York University law professor involved in litigation against the Police Department over its videotaping of protests like those surrounding the 2004 Republican National Convention. “But that’s in fact not the case,” he said. “The angle at which a camera is set, or the place where it’s set, and the time where it goes on, and goes off, tends to create an editorial space around the picture. But in fact people nevertheless continue to believe it’s the most reliable evidence of the scene.”
Protesters and onlookers have been policing the police with handheld cameras for some time, as one officer learned last summer after he shoved a man off his bike during a Critical Mass ride, and then lied about it in his paperwork. After the videotape surfaced, the officer was fired.
“Cops are paranoid about it,” said Mike Bosak, a retired sergeant who produces a daily e-mail digest of developments in New York law enforcement. “Lots of times some skell is fighting a cop tooth and nail, then a cop loses control, which is easy to do, and then you lose your temper and somebody videotapes you, and the next thing you know you’re losing your job.”
But with the department now encouraging New Yorkers to send cellphone photos and videos to its Crime Stoppers website, criminals are also more likely to be caught in the act. Yet interpreting the images is not always clear-cut. Arthur Eisenberg, legal director of the New York Civil Liberties Union, said the proliferation of cameras has lent a “Rashomon quality” to the testimony of witnesses to a crime. “The technology may help us come closer to an accurate picture of what happened, but inevitably, just as witnesses with their own eyes are viewing events from their own perspectives, so too are videographers,” he said.
The man knocked to the ground was identified by his lawyer as David Benzaquen, 25, a student at the New School. Mr. Benzaquen was charged with obstruction of governmental administration, resisting arrest and disorderly conduct, all misdemeanors. The police said he ran into an area that had been barricaded to exclude protesters and then “curled his body up into a ball,” to avoid being handcuffed, according to a criminal complaint against him. His lawyer, Yetta Kurland, said Mr. Benzaquen had done nothing to warrant the use of force. She also said he had been bloodied and bruised.

In a parallel story, Anne Eisenberg has an article in The New York Times about a new surveillance camera:

Standard surveillance cameras guard homes, businesses and public spaces around the clock, but they do the job imperfectly: fish-eye lenses can distort the image, and pan-and-tilt cameras may point the wrong way at a crucial moment.
Now a new, six-ounce security camera the size of a deck of cards uses a different approach to cover a 180-degree field of view. It employs fixed, inexpensive sensors that divvy up the surveillance job, and smart software that instantly puts their separate views together into one live streaming video.

The camera, the Digital Window D7, uses five 1.3 megapixel sensors just like the ones in camera phones, each aimed at its slice of the total view. Ingenious programs and a controller chip synchronize the five images as they are received, stitching them simultaneously into a panoramic stream that transmits at 15 frames a second.
To be released this year, and to be sold for about $800, the camera is a good example of what’s ahead in the field of computational photography, said Paul Worthington, a senior analyst at Future Image, a consulting firm in San Mateo, Calif. Computational photography, he said, draws on the power of computer software and processing to extend the capabilities of digital imaging.
The Digital Window security camera may be a bargain for consumers, Mr. Worthington said, in part because its powers come not from expensive components, but from software that makes use of off-the-shelf sensors. “Fabricating a single, large sensor is expensive,” he said. “Instead, this camera takes advantage of cheap sensors” and uses computer programming to unite their powers. “This is an impressive technology for surveillance,” he said. “It’s a more affordable way to get a high-quality, wide-angle image with no distortion.”
The camera is a product of Scallop Imaging, a division of the Tenebraex Corporation, an optical technology company in Boston. The name “scallop” was chosen because the rim of that bivalve’s shell is lined with a multitude of tiny eyes. “The scallop distributes the imaging task,” just as the Digital Window camera distributes its imaging job, said Peter Jones, president of Tenebraex. “We use lots of sensors and aggregate their images through the computer.”
The camera is part of a new generation of digital surveillance systems that can communicate over a computer network. Surveillance has traditionally been done by analog, closed-circuit TV systems, which still comprise the bulk of the security market. But because devices like Scallop’s, called Internet Protocol, or I.P., cameras, use computer networks, they can be accessed remotely and inexpensively through a Web browser.
“You can have 500 or 1,000 cameras all operating on one system,” said Alastair Hayfield, a market research analyst at IMS Research in Wellingborough, England. I.P. security products account for about fifteen percent of the total market for video surveillance, he said, and their sales are growing much faster than those of analog systems. “The I.P. devices are attracting a lot of attention from manufacturers looking to get into the market, which is becoming fairly substantial, with lots of long-term opportunities to make money,” he said.
At Scallop Imaging, Ellen Cargill, director of product development, said the Digital Window camera’s high-speed video interface grabbed data from the sensors, held it in its memory, then applied the calibration that created the final video stream. Users see the 180-degree view on their security monitor; they can also click in for a closer, high-resolution look at a fraction of the screen without losing the 180-degree view. “Typically when a camera zooms in, the full field of view is lost,” she said. “We are hoping to offer something different.”
Other companies, including Adobe in San Jose, California and Refocus Imaging in Mountain View, California, are developing systems that take advantage of computation photography techniques. Kevin Connor, vice president of product management for professional digital imaging at Adobe, says the company is experimenting with a combination of lenses and software to process enough data in one shot, for example, that the focus of the image can be changed after the shutter snaps, something that is now impossible. “We can capture the light rays in a scene and then later calculate a photo at any focal length,” he said. “You can refocus your shot after it is captured.” Using such a system, which produces data files that can be processed to create a focal point, a photo that is focused on a person’s face with the background showing only in a blur, for example, could be changed afterward so the fuzzy objects were now in focus.
At Refocus Imaging, Ren Ng, the chief executive, says the company is developing a next-generation digital camera with the same goal, changing the focus of an image after it is taken. “Computational photography is the future of imaging,” he said. “If you capture enough information, you can use software to perform the physical functions of the camera. You get possibilities that seem impossible, like refocusing.”