next up previous contents
Next: 4.2 Run CIV1: Up: CIV processing step by Previous: CIV processing step by   Contents

Check image quality:

This can be done with the uvmat programme, with the field selection image. Movies can be made to check a whole sequence. By simultaneously selecting a pair of images, for instance a and b, we can visually appreciate the feeling of continuity of particle patterns between the two images. Use the zoom option to get a closer view.

A very dense seeding of particles is the primary condition to check. Check also the luminosity and contrast of the images: fully black or saturated parts of the images must be avoided (by adjusting the luminosity and contrast of the camera). Look at the image histogram to check that.

A common source of difficulty is the occurence of fixed patterns due to illumination of the background. This effect can lead to zero velocity vectors. To avoid this effect, the first condition is to improve the quality of the illumination, using screens to avoid spurious reflexions. A poor water clarity can also contribute to illuminate the background by diffusion of the laser light. As a last recourse, one can substract the average to a sequence of images (e.g. by the option field_average in uvmat. This substraction removes the fixed background (but as a tradoff adds some noise).

Don't forget to check the calibration for distances: make images with a reference grid in the view field with the same camera position and objective lens. Then the calibration pixel to cm can be obtained. In uvmat, the pixel index of any point can be obtained with the mouse: first check the zoom option (left click to zoom and right click to unzoom), then press the select pushbutton to get the mouse cursor, and left click on the mouse. The pixel indices in x and y (as well as the image luminosity) appear in the text window at the upper right.

Adjust the angle of the camera such that walls are parallel to the x or y coordinates: this makes further processing much easier. A mask can be made to exclude some regions of the images in the processing, for instance to take into account the domain boundaries. The matlab programme makemask can be used for that purpose.


next up previous contents
Next: 4.2 Run CIV1: Up: CIV processing step by Previous: CIV processing step by   Contents
Joel Sommeria 2003-02-14