It started out as just a speck on a photograph of a man who threw a brick at truck driver Reginald Denny at Florence and Normandie avenues in the opening hours of the 1992 Los Angeles riots.
But when Leonid Rudin subjected it to a complicated computer algorithm and a slew of complex mathematical equations, that speck--originally less than 1/6,000th the size of the total photograph--was revealed to be a rose-shaped tattoo on the arm of the man, later identified in court as Damian Monroe Williams.
The tattoo was one of several details uncovered by Cognitech, Rudin's Santa Monica image processing firm, that helped place Williams and co-defendant Henry Keith Watson at the crime scene, said prosecutors, who won convictions of Williams, Watson and others involved in the beating of Denny and in assaults on several others at the intersection.
It was also the start of what Cognitech's co-chief executives, Rudin and Stanley Osher, believe could be a multibillion-dollar business in image enhancement for law enforcement.
"After the Reginald Denny case, we found out there was a market" for their uniquely sophisticated brand of image analysis, said Osher, who is a professor of mathematics at UCLA.
"We were just flexing our muscles, and suddenly our doors wouldn't stay closed and our phone wouldn't stop ringing," added Rudin, also Cognitech's director of research and development.
In the year since the Denny trial, Cognitech engineers--who had been entirely dependent on Defense Department research and analysis contracts until they applied their optics expertise to commercial markets--have analyzed dozens of videos and photographs for law enforcement agencies, insurance companies and private investigators.
At an average of cost of $5,000, though, their consulting services are not cheap. And the seven-person staff can only take on five cases a month.
So to meet demand in this emerging market and increase the company's current nearly $1 million in annual revenue, Rudin and Osher plan to sell their image enhancement technology in a do-it-yourself software package called "Police Video Workstation." They envision police detectives, insurance claim managers and others with desks equipped with powerful personal computers using Cognitech software. That software will be priced between $500 and $1,000.
"I think everybody would love to have it," said Beth McGee, a spokeswoman for the National Assn. of Police Organizations in Washington, D.C.
Video from surveillance cameras in convenience stores, parking garages and police patrol cars is increasingly being used as evidence by law enforcement and insurance companies, industry groups said. Cognitech's software clarifies those images, which are then used to find suspects and against those suspected of filing false claims.
For example, an automated teller machine camera happened to record someone in a car fleeing the scene of Manhattan Beach Police Officer Martin Ganz's killing last December.
"It was very far away and extremely fuzzy," Cognitech engineer Stephen Jensen said of the images recorded. "We sharpened it up and got information about the bumper, the wheels, the side molding, what color it was, where there were dents, etc."
Using Cognitech technology, police were able to identify the getaway car--a 1988 gray Daihatsu. When Ganz murder defendant Roger Hoan Brady was picked up earlier this month in Oregon, that car was in his possession.
Insurance companies have hired Cognitech to analyze videos of people suspected of making false disability claims. Industry groups say such claims cost insurers millions of dollars a year.
In one case, Rudin said, "we have a positive ID of a guy who's supposed to be paralyzed and he's playing football."
"Insurers are working hard to eliminate fraud, and they'll (take a serious look at buying) anything that is going to help them do that," said Bill Schroeder, vice president for claims of the Alliance of American Insurers in Schaumburg, Ill.
Traditional methods of enlarging photographs and video images rely on linear mathematics equations that smooth over background "static and noise," revealing general shapes but not the distinct features needed for identification. Cognitech's algorithms are based on far more complicated partial differential equations, which are used to design airplane wings and analyze supersonic booms. This nonlinear math can find the sharp edges in pictures, resulting in greater image clarity.
Terry Davis, an investment analyst with Ventana Growth Fund in Irvine, said Cognitech is the only company using this type of technology for image processing.
Rudin and Osher say that is because it is a cutting-edge approach developed for military use.
"This technology used to be directed mostly toward scientific pursuits, but with the decline of aerospace industry, more and more attention is being given to law enforcement and public safety issues," said Terrance Montonye, technical director of the International Society of Optical Engineering, which for the first time included a session on law enforcement technology at its annual meeting in July.
About 10% of the country's 15,000 to 17,000 law-enforcement agencies are large enough to easily afford the nearly $10,000 it will cost to buy the Cognitech software and the computer equipment to run it, said Craig Fraser, associate director of Police Executive Research, a Washington, D.C. think tank. He said more departments will enter the market as the price of computer hardware drops.
Rudin said Cognitech plans to lend its Police Video Workstations to the Los Angeles Police Department and nine other departments around the country early next year for a six-month trial. He said he expects to bring the product to market at the end of the year.
That is not soon enough for Faye Arfa, a lawyer who successfully defended a client against charges of attempted murder and robbery with the help of Cognitech imaging analysis. Its analysis of video from a convenience store camera contradicted the account of an eyewitness against Arfa's client.
"It's a significant breakthrough and a very useful tool," Arfa said. "It would be a real asset to be able to have this technology available on an everyday basis."