Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!usc!jarthur!nntp-server.caltech.edu!lo-fan.caltech.edu!sns From: sns@lo-fan.caltech.edu (Sam Southard Jr.) Newsgroups: comp.compression Subject: Re: Atronomical data compression Message-ID: <1991Mar25.224706.7129@nntp-server.caltech.edu> Date: 25 Mar 91 22:47:06 GMT References: <1991Mar23.013557.28151@nntp-server.caltech.edu> Sender: news@nntp-server.caltech.edu Reply-To: sns@lo-fan.caltech.edu (Sam Southard Jr.) Organization: Caltech Astronomy Department Lines: 35 In article , philip@beeblebrox.dle.dg.com (Philip Gladstone) writes: |> >>>>> On 23 Mar 91 01:35:57 GMT, sns@lo-fan.caltech.edu (Sam Southard Jr.) said: |> |> The first question that comes to mind is whether you want a lossless |> scheme or a lossy scheme? |> |> If you want lossless compression, then I must ask you why? You already |> admit that there is noise in the input signal and I bet you get noise |> introduced from atmospheric turbulence. This question has been asked by many people who have sent me responses to my original message (BTW: Thanks!). I apologize for not answering it in my inital posting - since I was so involved with the thing the answer was so obvious to me thast I didn't even bother addressing the issue. The answer is, I need lossless compression. This code is going in the actual instrument. Astronomers will not be happy if they were informed that they will get anything other than the actual data, bit for bit. The decision which was made (I agree with it) was that the astronomer should decide what to do with the data. One of the reasons for this is that these instruments are often used to look at very faint light sources and features which are quite often fairly close to the noise level. It would be extremely difficult for anyone but the individual astronomer to decide if a particular small delta somewhere was significant. Imagine what an astronomer would do to me if he found out that the spectral line he based his thesis on in fact might have been merely noise introduced by a compression algorithm! |> If you want lossy compression, then you have to decide how much loss |> you are prepared to accept, In general the lossy alogorithms do much |> better at compressing pictures than the lossless ones. The improvement |> can be a factor of 10 better. Unfortunately, they are completely unacceptable in this case.