No abstract

Before being applied in oil fields, it is necessary to evaluate the quality of the fracturing proppants, for which sphericity and roundness are the two important parameters. In this paper, we propose a new approach to the sphericity and roundness measurement of proppants based on an image processing method. First, the OTSU method is introduced to binarize proppant image, and the statistical analysis is performed to divide the image into proppant regions and non-proppant regions. Second, after filling small holes, edge curves of all proppants were obtained by the Canny operator. Third, for all 20 images in the standard template, according to their corresponding sphericity and ratios between the radius of maximum inscribed circle and the radius of the minimum circumscribed circle, a fitting curve with maximum coefficient of determination was obtained, which could determine any proppant's sphericity. Fourth, the edge curve of any proppant was divided into several segments and curve fitting method was used to obtain a curve function for each segment. Finally, the roundness of any proppant was obtained based on curvature radius of all points in the edge curve and radius of its minimum circumscribed circle. Actual proppant's evaluation experiments showed that our approach was closest to the artificial determination using a standard template in terms of precision. INDEX TERMS Sphericity, roundness, proppant, image processing.

We investigate the problem of training an oil spill detection model with small data. Most existing machine-learning-based oil spill detection models rely heavily on big training data. However, big amounts of oil spill observation data are difficult to access in practice. To address this limitation, we developed a multiscale conditional adversarial network (MCAN) consisting of a series of adversarial networks at multiple scales. The adversarial network at each scale consists of a generator and a discriminator. The generator aims at producing an oil spill detection map as authentically as possible. The discriminator tries its best to distinguish the generated detection map from the reference data. The training procedure of MCAN commences at the coarsest scale and operates in a coarse-to-fine fashion. The multiscale architecture comprehensively captures both global and local oil spill characteristics, and the adversarial training enhances the model’s representational power via the generated data. These properties empower the MCAN with the capability of learning with small oil spill observation data. Empirical evaluations validate that our MCAN trained with four oil spill observation images accurately detects oil spills in new images.

No abstract

No abstract