the zoom factor in most cases, just skip reading some pixels but they are still sensed out by the camera. So the same overall duration is normal with such a method. Specially for 480xX, the full resolution is used but with windowing. additionally requiring 480 pixel on a line with a zoom of 8 make no sense. The image is 480X640 so if you want 480 pixels the only zoom that make sense is 1. Not sure about the jitter, maybe some counter goes out of scope or something else. Maybe other bugs are still in, like counters or buffers limited to 256.
the driver might be slightly different to solve that issue, but it was some time ago ;-).

good luck, and let us know if you solved the issue.

On 20:59, Duarte Dias wrote:
Hello Giles,

I did the time measurments for both microcontroler acquisition and bluetooth comunication to the PC.

For the BT transmission duration, I've noticed that the more lines I extracted, the more bytes I had to transmit, so the transmision time increased as expected.

For the microcontroler acquisition duration, I was expecting also an increase with the number of lines taken, since more lines needed to be acquired. Instead I get more or less a constant value arround 400ms, for evry camera configuration, with a huje jitter value (standard deviation of the captures is arround 100-150ms).

I went to po3030k.h where it is written the frame rates for some camera configurations. Mine is not there ( GRAY (1-8)x480 with zoom of 1,2,4,8 ) but I was expecting it be arround the maximum frame rate (8 fps). Instead I'm getting arround 2 fps with huje jitter.

I want to take out the Jitter of the captures and obtain the maximum frame rate that the configurations I am using allow. What should be the durations that I should obtain for microcontroler acquisition with my camera configurations?


2011/7/21 Gilles Caprari <g.capr...@ieee.org <mailto:g.capr...@ieee.org>>

    Dear Duarte,

    this example might be an inspiration

    usign the zoom basically skips pixels and lines and takes only the
    one usefull. Not sure if you mean acquisition from the
    microcontroller or also on the PC via bluetooth (BT). In that
    later case, most of the time required by the BT communication.


    On 20:59, Duarte Dias wrote:

    I am using  the e-puck 1176 (VGA camera 480x640 - rotated camera).

    I am capturing some full (480 pixels) horizontal lines (1 to 8
    lines) applying the zoom factor of 1, 2, 4 and 8 (on the x
    component which in this epuck camera is the vertical component).

    I assume that if I am capturing 4 horizontal lines with zoom 2
    its the same of capturing 8 lines with zoom 1 (and so one). The
    zoom just does the line averaging within groups of 2 lines.

    When I capture the images I am using a timer (TIMER1) to quantify
    the capture duration.

    After my experiments from capturing 1 - 8 lines with zoom 1,2,4,8
    I have concluded that the capture time is approximatly constant
    (between 350 and 450), no matter the number of lines we take and
    the zoom we do.

    I have also witnessed a large Jitter between frame captures (the
    standard deviation of the frame capture times is arround 150ms!).

    I was expecting a rise of time when I extracted more lines and no
    Jitter should exist. Could someone tell me what I could be the
    cause of this behaviour (expecially the Jitter behaviour).

    Duarte Dias

-- ------------------------
    Dr. Gilles Caprari
    GCtronic Sagl, Electronics&  Mechatronics
    web:www.gctronic.com  <http://www.gctronic.com>  E-mail:g.capr...@ieee.org  
    Post: Via Cassarinetta 2, CH-6900 Lugano, Switzerland
    Tel: +41 91 630 2149 Fax: +41 43 556 8006
    Mobile: +41 78 601 1113 skype: gillescaprari

Dr. Gilles Caprari
GCtronic Sagl, Electronics&  Mechatronics
web: www.gctronic.com E-mail: g.capr...@ieee.org
Post: Via Cassarinetta 2, CH-6900 Lugano, Switzerland
Tel: +41 91 630 2149 Fax: +41 43 556 8006
Mobile: +41 78 601 1113 skype: gillescaprari

E-puck-user mailing list

Reply via email to