Computer vision is a rapidly growing field devoted to analyzing, modifying and high level understanding of images. Beside its wide range of use, computer vision is computationally expensive, thus real-time processing became possible only due to the introduction of strong processing units and architectures via the advancements in the semiconductor industry. These advancements, through the cost-efficient CMOS sensors, also enabled the digital cameras to become widespread, which, combined with today’s level of computational capacity, raises the need for computer vision based applications and helps with their spread.
Beside everyday applications, many instances can be found for the industrial and scientific use of computer vision due to its precision, cleanliness, and its ability to simplify and automate tasks. The inspection of two-phased systems (not miscible liquids or gas-liquid mixtures) is a good example for the potential of new methods, where, through locating and processing the phase boundaries, the volumetric flow rate may be determined.
By the means of gradual improvement of said flow rate measuring application and its test environment, I achieved a system that is able to detect the different phases in gas-liquid solutions and after their tracking, the bubbles’ volume may be determined.