An automatic image processing technique for analyzing the laser images of two-phase (oil and water) pipeline flow is proposed. The main steps of this technique include a new automatic thresholding method for conversion from gray-level to binary images, binary smoothing, labeling, examining, and calculating. The technique has proven reliable to automatically analyze the images of this two-phase-flow research. By random searching, bright water droplets (droplets with high gray level) are detected. Based on the characteristics of the bright droplets, the threshold is determined by the new thresholding method, called cross-edge-line thresholding. Experimental observation shows that this thresholding method can be used successfully for two-phase laser images. Binary smoothing is used to smooth the boundary of the extracted droplets. Labeling is then used to distinguish each separate droplet. In the examining step, elimination of faint droplets and nondroplets and multiplication by a possibility factor to account for droplets only partially immersed in the control volume are done to improve the performance of the technique.
Emil L. Hanzevack,
"New Technique To Automatically Analyze Laser Digital Images Of Two-Phase Pipeline Flow," Optical Engineering 27(7), 277528 (1 July 1988). http://dx.doi.org/10.1117/12.7976718