Edge Detection and Machine Learning Approach to Identify Flow Structures on Schlieren and Shadowgraph ImagesстатьяИсследовательская статьяЭлектронная публикация
Дата последнего поиска статьи во внешних источниках: 1 декабря 2021 г.
Аннотация:Schlieren, shadowgraph and other types of refraction-based techniqueshave been often used to study gas flow structures. They can capture strong density gradients, such as shock waves. Shock wave detection is a very importanttask in analyzing unsteady gas flows. High-speed imaging systems, includinghigh-speed cameras, are widely used to record large arrays of shadowgraph images. To process large datasets of the high-speed shadowgraph images and automatically detect shock waves, convective plumes and other gas flow structures,two computer software systems based on the edge detection and machine learningwith convolutional neural networks (CNN) were developed. The edge-detectionsoftware utilizes image filtering, noise removing, background image subtractionin the frequency domain and edge detection based on the Canny algorithm. Themachine learning software is based on CNN. We developed two neural networksworking together. The first one classifies the image dataset and finds images withshock waves. The other CNN solves the regression task and defines shock waveposition (single number) based on image pixels tensor (3-D array of numbers) foreach image. The supervised learning code based on example input-output pairswas developed to train models. It was shown, that the machine learning approachgives better results in shock wave detection accuracy, especially for low-qualityimages with a strong noise level. Software system for automated shadowgraphimages processing and x-t curves of the shock wave and convective plume movement plotting was developed.