Ron, Art, & Aus,
I've just spent 45 mins typing in some detail. It's all lost!
Ok, I'll try again, but it might be briefer.
First, thankyou to all of you. I had neglected the video. From what you say, I shouldn't. I still think though that, for a monitoring system, single-shots would be better.
The environment is harsh. The fewer moving parts the better. The observers have many other duties, their primary purpose is to monitor the vessels interaction with marine mammals. They couldn't edit video tapes, and managing 30 to 80 hours of video would be an encumbrance for them - they have little space. I can't see that amount of footage being transmitted.
We are going to have many samples, so we can tolerate imprecision in our measurements of those samples. We cannot tolerate bias though. An error of even 10% can be tolerated in each sample so long as the errors are not consistent - as long as we don't tend over or under-estimate sizes. Adding all the errors together leads to an overall error that gets smaller as a proportion of the total.
A problem is that the fish probably come aboard the boat and go down the ramp in an order that depends upon many things, so the time at which photos are taken is important and could create bias. We need to be able to recreate the size and species composition of fish ion the net from the samples that we take. We could take the photos at random - not as easy as it seems - but given that there is order, we could probably get a more accurate picture of what was in the net for a fixed number of photos if we paid careful attention to when the photos were taken.
I hope there is a camera that could be programmed to, say, take a picture every twenty seconds for five minutes, then every minute.
Having said that, I'm beginning to think that we might have to stick with video. Certainly video could be scanned to find the ends of each set, then frames could be taken from particular periods. I don't think it would be necessary to digitize these. there are programs available that, with the image displayed on the sceen and calibrated, would allow fish to be measured with two clicks of a mouse.
Anyway I'll end here to avoid further loss. |
