Ultrahigh-speed ray image recognition method based on photoelectric detector

Abstract

The invention discloses an ultrahigh-speed ray image recognition method based on a photoelectric detector. The ultrahigh-speed ray image recognition method comprises the steps of obtaining a plurality of input images after X-rays penetrate through an object; building a neural network, and training and verifying the neural network by using the plurality of input images to obtain a trained and verified neural network which comprises weight information and can output category information of the input images; weight information is extracted, a bias processor of the photoelectric detector corresponding to the multiple input images is constructed, and the bias processor can correspondingly adjust output pulse photoelectric signals of the photoelectric detector to achieve the weight calculation process of the neural network; according to the method, the bias device is utilized to perform different bias adjustment on the voltage of the photoelectric detector, so that the adjusted photoelectric detector performs corresponding adjustment on the detected pulse photoelectric signal generated by each pixel, then connection is performed according to the corresponding relation in the neural network, and the identification result of the X-ray image of the target object can be obtained.

Description

technical field [0001] The present application relates to the technical field of on-line monitoring of laser processing, in particular, to an ultra-high-speed ray image recognition method based on a photodetector. Background technique [0002] Laser thermal processing refers to the use of the thermal effect generated by the laser beam projected onto the surface of the material to complete the processing process, including laser welding, laser engraving and cutting, surface modification, laser laser marking, laser drilling and micromachining. In the process of laser processing, due to the difference in processing materials and processing requirements, it is often necessary to adjust the process parameters in real time to improve the processing quality. However, the properties of materials are various, and the internal structure is also very different, and it is difficult to obtain the internal structure of the processing through real-time monitoring during the processing proc...

AI Technical Summary

Problems solved by technology

However, this method is susceptible to interference from the external environment, resulting in deviations in results, and secondly, due to the different properties of materials, it is easy to cause distortion

Moreover, the current laser processing online monitoring technology mainly monitors surface phenomena, and lacks effective monitoring of defects in the depth direction, and defects in the depth direction are the key factors that cause material failure such as cracks.

Images