Processing of images containing overlapping particles

• Main Authors: BRAUNGER NEIL, CONWAY PATRICK, VALLOTON, PETER
• Format: Patent
• Language: Chinese; English
• Published: 03.02.2023
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; PHYSICS

The present disclosure provides a computer-implemented method of generating training data for training a machine learning model for generating a segmentation mask of an image containing overlapping particles. Training data is generated from sparse particle images that do not contain overlaps. If the particles can be clearly confirmed, the mask that generates non-overlapping particles is not typically a problem; simple methods, such as thresholding, in many cases have made available masks. Sparse images may then be combined to images containing artificial overlays. The same approach may also be employed for masks that produce large amounts of training data, as many combinations may be created from only small sets of images. The method is simple and efficient and can be adapted to many domains, such as by adding style migration to the generated image or by including additional augmentation steps. 本公开提供一种生成用于训练机器学习模型的训练数据的计算机实现方法，所述机器学习模型用于生成包含重叠粒子的图像的分割掩膜。从不包含重叠的稀疏粒子图像生成训练数据。如果粒子可以被清楚地确认，则生成非重叠粒子的掩膜通常不是问题；在许多情况