Xref: utzoo sci.electronics:2201 rec.ham-radio:3984 rec.autos:7200 Path: utzoo!mnetor!uunet!husc6!mit-eddie!ll-xn!ames!aurora!labrea!glacier!jbn From: jbn@glacier.STANFORD.EDU (John B. Nagle) Newsgroups: sci.electronics,rec.ham-radio,rec.autos Subject: Re: Robocop spotted (Photo radar enforcement) Message-ID: <17308@glacier.STANFORD.EDU> Date: 11 Feb 88 05:37:17 GMT References: <602@anasaz.UUCP> <9504@steinmetz.steinmetz.UUCP> Reply-To: jbn@glacier.UUCP (John B. Nagle) Organization: Stanford University Lines: 29 Gadgets like this, although rare in the US, have been around for years in Germany and Japan. But the radar/lidar approach really ought to be replaced with a motion vision system which matched the images from frame to frame and measured speed in that way. The major problem is that television resolution isn't good enough to read a licence plate given a field of view that covers an entire road. But once HDTV cameras become generally available, this problem should go away. Even without HDTV technology, it may be possible to make this work, although it will probably take one camera per lane. The major advantage of the video approach is that misidentification of the vehicle can be much reduced, if not eliminated. With a videotape as a check, there's an objective but human-checkable record to be examined in court when necessary. This should make it a lot harder to be falsely identified as the speeding car, while making it much easier to get convictions in court. Digesting the data at TV rates is hard, but not impossible. Signal processing chips that can convolve images in real-time are starting to become widely available. (Any DSP5600 users out there, by the way?) Note that the only exposed component of the system is the camera, and cameras are not only cheap, but tiny. (Pulnix has some very nice cameras about 1" x 1" x4"; these are becoming popular with the robotics people, especially when you want a camera near the end of an arm.) There could be a nice product opportunity here for someone. John Nagle