| CPC G06F 18/24 (2023.01) [G06N 3/045 (2023.01); G06N 3/088 (2013.01); G16H 30/20 (2018.01)] | 20 Claims |
|
1. A device, comprising:
one or more memories; and
one or more processors, communicatively coupled to the one or more memories, to:
receive a training data set for training a laser speckle reduction model,
wherein the training data set includes a set of image pairs of training objects, and
wherein an image pair, of the set of image pairs, includes a first image, of a training object of the training objects, that includes laser speckle and a second image, of the training object, that includes less laser speckle than the first image;
train, using a deep learning technique, the laser speckle reduction model based on the training data set,
wherein the laser speckle reduction model comprises a generator model to generate a reduced laser speckle image and a discriminator model to train the generator model;
receive, after training the laser speckle reduction model, a coherent energy illuminated image, of a particular object not included in the training objects, that includes laser speckle;
process, using the laser speckle reduction model, the coherent energy illuminated image to generate a laser speckle-reduced image,
wherein processing the coherent energy illuminated image comprises processing the coherent energy illuminated image in connection with an incoherent energy illuminated image; and
provide an output connected with the laser speckle-reduced image and the incoherent energy illuminated image.
|
|
14. A non-transitory computer-readable medium storing instructions, the instructions comprising:
one or more instructions that, when executed by one or more processors, cause the one or more processors to:
receive a training data set for training a laser noise reduction model,
wherein the training data set includes a set of image pairs of training objects, and
wherein an image pair, of the set of image pairs, includes a first image, of a training object of the training objects, that includes laser noise and a second image, of the training object, that includes less laser noise than the first image;
train, using a deep learning technique, the laser noise reduction model based on the training data set,
wherein the laser noise reduction model comprises a generator model to generate a reduced laser noise image and a discriminator model to train the generator model;
provide the laser noise reduction model based on training the laser noise reduction model;
receive, after training the laser noise reduction model, a coherent energy illuminated image, of a particular object not included in the training data set, that includes laser noise;
process, using the laser noise reduction model, the coherent energy illuminated image to generate a laser noise-reduced image;
provide the laser noise-reduced image as output;
perform image processing on the laser noise-reduced image to identify a characteristic of the laser noise-reduced image; and
provide additional output associated with the characteristic of the laser noise-reduced image.
|
|
17. A method, comprising:
receiving, by a device, a coherent energy illuminated image, of a particular object, that includes laser speckle;
processing, by the device and using a laser speckle reduction model, the coherent energy illuminated image to generate a laser speckle-reduced image,
wherein processing the coherent energy illuminated imagea comprises processing the coherent energy illuminated image in connection with an incoherent energy illuminated image; and
providing, by the device, an output connected with the laser speckle-reduced image and the incoherent energy illuminated image to permit diagnostics based on the laser speckle-reduced image.
|