23-06-2023 дата публикации
Номер: CN116275388A
Принадлежит:
The invention provides a welding laser visual image segmentation method, and belongs to the technical field of image segmentation. Comprising the following steps: S1, acquiring a laser stripe image by using a laser vision system, and preprocessing the laser stripe image; s2, building a neural network for laser visual image segmentation; s3, training the laser vision segmentation neural network, inputting the processed laser stripe image into the laser vision image segmentation neural network, and outputting a grayscale image with a grayscale value range of 0-255; and S4, training the neural network by using a back propagation algorithm, inputting the laser stripe image into the neural network for laser visual image segmentation, and outputting a segmentation result. The technical problems that in the prior art, a traditional image processing algorithm is poor in anti-interference capacity, and welding seam tracking errors are large under the interference conditions of welding reflection ...
Подробнее