Diffusing, Diverging, Attenuating Rays

Diffusing, Diverging, Attenuating Rays

This option allows for crossing light rays in a relatively simple manner. The correction works reasonably well with light from a large apperture source (such as a linear halogen photographic lamp), but is not as successful with light from a small source (such as a laser or arc lamp) where the light rays do not cross (in which case [D Diverging, attenuating light rays] is more appropriate).

In principle the intensity Pr(x,y,é) of a light ray at angle é to the x axis decays as

       dPr = -A(C) Pr ds,                                           (1)

where A(C) is the attenuation rate as a function of concentration C=C(x,y) for which the dye is a nominal marker (and hence A=A(C)=A(x,y)), and ds=dy/siné is the incrimental distance a ray of light travels. The overall intensity of light at a particular point, defined by

                ôã
       P(x,y) = ³  Pr(x,y,é) dé,                                    (2)
                õ0

behaves as

       dP   ôã dPr
       ÄÄ = ³  ÄÄÄ dé
       dy   õ0 dy

            ôã dPr
          = ³  ÄÄÄ cosecé dé
            õ0 ds

              ôã
          = - ³  A(C) Pr cosecé dé.                                 (3)
              õ0

For weak concentrations we may take

                                                                    (4)
       A(C) = a C,

where a is a constant (the concentration C is independent of angle) so that

       dP         ôã
       ÄÄ = - a C ³  Pr cosecé dé.                                  (5)
       dy         õ0

The finite difference approximation with incriments ëI and ëy allows this equation to be rewritten as

                     ôã
       ëP = - ëy a C ³  Pr cosecé dé,                               (6)
                     õ0

in which we may replace cosecé dé with an integral in x to give

                   ôx+ì                     dx'
       ëP = ëy a C ³     Pr(x',y-ëy,é) ÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄ.            (7)
                   õx-ì               û[(x-x')ý + ëyý]

Without loss of generality we may write

       Pr(x,y,é) = P(x,y) T(x,y,coté),                              (8)

where T(x,y,coté) gives the relative amplitude of the different rays passing through a point x,y at an angle é and from the definition of P(x,y)

           ôx+ì
           ³  T(x,y,x'/ëy) dx' = 1.                                 (9)
           õx-ì

This may be substituted into equation (7) to yield

                   ôx+ì                                        dx'
       ëP = ëy a C ³     P(x',y-ëy) T(x',y-ëy,(x'-x)/ëy)  ÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄ.
                   õx-ì                                   û[(x-x')ý + ëyý]
                                                                    (10)

To simplify the image correction problem we shall assume that T/û[(x-x')ý+ëyý] can be split into filter and amplitude components. The filter component F(x,y,x') depends only on the geometric properties of the light sheets and may be approximated in a simple manner by a low pass filter combined with a translation allowing for the nominal ray direction. This filter component caters for the divergence and crossing of the light rays. For brevity we shall write this filter operation as

                  ôx+ì
       f(P,x,y) = ³     P(x',y-ëy) F(x,y,x') dx'.                   (11)
                  õx-ì

The amplitude component G(x,y) parameterises the decay of the illumination due to divergence and absorbtion of the medium (excluding the fluorescent dye). This amplitude component will be obtained from a calibration image of uniform dye concentration. Using the above assumption equation (10)

       ëP(x,y) = ëy a C(x,y) G(x,y) f(P,x,y).                       (12)

Suppose P0, C0 are the intensity and concentration fields for a calibration image of uniform concentration. From this calibration

       G(x,y) = ëP0(x,y) / [ëy a C0 f(P0,x,y)],                     (13)

so that (12) may be written as

                 C(x,y) f(P,x,y)
       ëP(x,y) = ÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄ ëP0.                               (14)
                   C0 f(P0,x,y)

Following the linear attenuation approximation, we assume the observed intensity field p(x,y) is related to the concentration and illumination by

       p(x,y) = à C(x,y) P(x,y).                                    (15)

Substituting this into (14) we may see that

                    f(p,x,y)
       ëP(x,y) = ÄÄÄÄÄÄÄÄÄÄÄÄÄÄ ëp0,                                (16)
                 à C0 f(p0,x,y)

which provides the method used to determine the intensity of the light sheet in this option. Finally, once P(x,y) is determined by stepping (16) through an image, we may determine the concentration from

       C(x,y) = p(x,y)/[à P(x,y)]                                   (17)

A knowledge of the fluorescing efficiency à is not required unless the actual intensities within the sheet are required. The actual concentration of the fluorescent dye constituting a nominal unit concentration (as defined implicitly by C0) may be chosen to optimise the quality of the images.

NOTE: Before applying this correction, the image intensities should be mapped onto a linear intensity scale passing through zero.

Correct LIF image in which buffer ?

This entry specifies the buffer containing the LIF image to be corrected for attenuation in the light sheet.

Store (first) corrected LIF image in which buffer ?

This entry specifies where the final corrected LIF image is saved. This buffer may be the same as that containing the original image. If more than one buffer was specified in the source list, then the second and subsequent images are stored in sequence starting at this buffer.

Store illuminating intensity in which buffer ?

During the correction process the intensity of the illuminating light sheet is determined. This entry specifies the buffer into which it is written. If more than one buffer was specified in the source list, then this buffer will be reused for each subsequent image.

Constant concentration LIF image (used for calibration) is in which buffer ?

To calibrate the concentrations and determine the diffusion and attenuation of the light sheet it is necessary to capture an image of the set up containing a constant dye concentration. This entry specifies the buffer in which such an image may be found. This image is also used to determine th intensity of the light sheet where it enters the flow. The intensity of this sheet is normally set to P/C0, where P is the intensity of the constant concentration image at the entry point, and C0 the concentration (see next entry). However, if this would result in saturation of the image, then the intensity of the sheet will be rescaled to avoid this (a message to this effect is also produced). Note that this image should have been previously corrected for the camera/VTR intensity transfer function.

Concentration for calibration image ?

This entry specifies the concentration associated with the calibration image stored in the buffer indicated by the previous entry. Concentrations should be in the range 0 (no dye) to 1 (maximum dye concentration)..

Correct in Window or whole Screen ?

The image may be corrected in either a window (<W>) or the entire frame buffer (<S>). Note that it is essential that the whole height and width of the window contain a valid LIF image otherwise the fitting of the light sheet will not function correctly.

Window Selection: Region to correct for light sheet attenuation

If the user choses to correct for light sheet attenuation in a window, then the Window Selection submenu is produced to aid the window specification process. Additional information on window specification is available through the [H Help] option within the submenu.

Light rays entering flow at Bottom, Top, Left or Right ?

The direction in which the light rays are travelling must be specified in order to calculate the correction. At present DigImage requires the light rays to be (approximately) parallel with the pixel coordinate system, and the intensity uniform at the boundary of the screen or window through which they enter. Care should be used when applying this algorithm to situations where the light rays are not parallel.

Use which buffer for defining ray directions and spread ?

This entry allows selection of a suitable buffer for defining the orientation and spread (diffusion) of the light rays entering the field of view. Often the ray directions are most easily established by masking of a number of narrow strips in the light sheet such that a grid of lines is projected into the tank containing a uniform fluorescein concentration. An image of these lines will then clearly define the orientation of the light rays. Note that with a good light sheet the rays should not normally be obvious in a uniform dye concentration!

Cursor Positioning: Entry point for ray

    The Cursor submenu is produced to allow each entry point for a diverging
    light ray (such as that marked by the shadow cast by some slide or mask)
    to be specified. The point chosen should be at or near the boundary of the
    image through which the light ray enters. For assistance on the use of the
    cursor, consult [H Help] within the cursor submenu.

Move triangle: <left>-, -<right>, Condense, Enlarge, change Step size, Manual or Quit ? .52 Move triangle: <up>, <down>, Condense, Enlarge, change Step size, Manual or Quit ?

    This selection allows the orientation and spread of the divergence
    triangle to be changed. The left and right or up and down arrow keys move
    the triangle in the appropriate direction. <C> reduces the width of the
    triangle while <E> enlarges the width.  <S> may be used to
    toggle between steps of one and sixteen pixels at a time. Once the
    triangle is positioned, <Q> continues execution. From a command file it is
    more convenient to enter the rays manually using <M> which prompts for the
    angle to the nominal ray direction and the spread angle.

{If ray direction entered manually}

Angle of rays to nominal direction ?

This prompt is produced only when the ray directions are entered manually rather than interactively. The mean ray angle is specified relative to the nominal ray direction (the nominal ray direction is normal to the side of the image through which the ray enters). The angle should be specified in degrees.

{If ray direction entered manually}

Spread angle for rays ?

This prompt is produced only when the ray directions are entered manually rather than interactively. The spread angle is the angle between the two sides of the triangle. The angle should be specified in degrees.

Specify another ray triangle ?

    Between four and ten ray triangles may be specified. They will
    subsequently be used to generate a least squares mapping of the spread of
    the light rays. The first three times this prompt is produced only the
    "yes" answer <Y> will be accepted, thereafter either yes <Y> or no <N> is
    valid.

Message:
Determining light sheet intensity...

Correcting image for light sheet...


Parent menu

Main Menu

DigImage User Documentation


Stuart Dalziel, last page update: 19 February 1996