Shop OBEX P1 Docs P2 Docs Learn Events
Creating a simple VEDR: Video Event Data Recorder — Parallax Forums

Creating a simple VEDR: Video Event Data Recorder

DirkKDirkK Posts: 3
edited 2007-12-30 19:31 in Propeller 1
Greetings,
Does anybody know of any resources that would guide me in using Parallax products to create a simple Video Event Data Recorder?·
In other words, I'd like to be able to write a video stream to RAM and, using a manual trigger or event, save the last [noparse][[/noparse]x] seconds to flash.· Yes, you may recognize this as the description for a 'black box'.·
For example: if a robot is wandering around and an event occurs, I'd like to be able to view the video stream just prior to the event.· I could simply record everything to flash, but that would require too much flash and the constant writing would shorten·the flash·life relatively quickly.

From the reading I've done so far, it looks like I might need additional chips to handle the encoding of the video stream.
Suggestions?

- Dirk

Comments

  • Mike GreenMike Green Posts: 23,101
    edited 2007-12-29 20:56
    The main problem you would have is providing enough RAM. Video requires an immense amount of storage unless you significantly limit the resolution and frame rate. There are compression algorithms, but they're very compute intensive and commonly require at least two frames of storage. You would certainly need external memory at the very least. You would need a high speed ADC and sync separator and a variety of support logic to make this all work fast enough with the number of I/O pins available.
  • rjo_rjo_ Posts: 1,825
    edited 2007-12-30 13:27
    I am assuming that the event that you are looking for is a simple sensor event.

    You might want to take a look at Bean's overlay board (Hitt Consulting)... the board overlays graphics of your choosing onto an ordinary video stream... in theory you could use Bean's overlay technique to encode a binary time stamp and event marker right onto your video... then you just need a video recorder that you can control through one of those serial protocols that everyone eats for breakfast around here.

    You could feed this stamped video into a computer real time and use ImageJ to decode the time references (public domain from NIH) and obtain image derived data.

    When you want to look at the data prior to an event you will be parsing through one or two lines of video... which the Prop can easily do.

    For A/D have a look at ViewPort.
  • DirkKDirkK Posts: 3
    edited 2007-12-30 19:31
    Thanks for the feedback.
    Yes, I expect to have to come up with more RAM.· However, I do plan to plan to limit the resolution and frame rate as you suggest (320 x 240 at perhaps 8fps).· Also, I only plan to only capture·about 10 seconds of video per event.· Given that, it seems like even 16mb of additional·RAM would work.
    It·seems like the real challenge is finding and integrating the ADC (maybe trivial for somebody who has already familiar with it, but a long road for a new comer).
    ·
Sign In or Register to comment.