Optical micro-bubble sensor
For the measurement of very small gas bubbles in fluids we developed a special optical sensor based on the principle of videoendoscopic flow observation combined with a sophisticated image processing software for detection and measurement of micro-bubbles. Today, particle and gas bubble detection and measurement is a typical domain of numerous optical sensors based on light scattering. However, commercial particle detectors and phase-Doppler particle anemometer devices are bulky and expensive and often require multiple optical access to the examined fluid volume. Such devices have been developed for a broader range of particle measurement problems, comprising particles of different shapes, materials, refractive indicees, and for various flow types. We use a conceptionally simpler and thus less expensive measurement approach based on optical transillumination and image processing. This method gives highest accuracy of measurement even for larger gas fractions or more complex particle shapes.
The principle design of the optical sensor is shown in the right figure. The sensor consists of a metal housing containing the camera and the control electronics and a 150 mm long tubular shaft enclosing an endoscopic optics. At the tip of the tubular shaft there is a slit of 1.5 mm width in which the flow is optically observed. A high-power LED light source at the distal end of the shaft produces the light that transilluminates the slit. On the other side of the slit an endoscopic optics transfers the image of the 2.85 mm x 2.15 mm wide field of view to the camera which resides in the sensor housing. The CCD camera can generate 30 images per second with an electronic shutter exposure down to 10 µs. To be able to observe gas bubbles at flow velocities up to 5 m/s the LED can be flashed in synchronization with the camera exposure for down to 1 µs time duration. Due to the short exposure time the images are to some degree noisy having an SNR of approximately 10. A digital image as acquired by the framegrabber has a dimension of 640 x 480 pixels and a gray value depth of 8 bit.
With the sensor it is possible to acquire a series of several hundred images within a given time regime. After acquisition the images are processed in a first simple image algorithm step to produce binary images that represent homogeneously white background and black bubble shadows. Therefore each image is properly processed by dark subtraction, shading correction, background extraction by a boundary fill algorithm and bubble filling by a region fill algorithm. After these steps image pixels are labelled according to their affiliation.
In a second step the bubbles that are present in the image are digitally extracted and measured. Therefore we can either use a straight forward algorithm that measures the area of each coherent shadow region and computes an equivalent bubble diameter from the region area. A more sophisticated algorithm, however, can be used to identify circular bubbles within the black regions. This algorithm is based on the Hough transform for circles - a versatile and highly accurate pattern analysis approach, that is frequently used to extract primitive or complex shapes from noisy images. With this algorithm it is possible to dissolve so called bubble clusters, i. e. agglomeration of many bubbles in one coherent black image region, into single bubbles. For higher gas fractions or a larger slit volume this gives more accurate bubble statistics and gas fraction values.
The optical sensor can be applied in many areas of industrial measurement and basic research where small gas bubbles occur due to local vapourisation or gas production on a miniature scale. Examples are
- chemical reactions under varying hydrodynamic pressures,
- fermentation processes,
- boiling as well as
- gas dissolution and dispersion in fluids.