Research on Fresnel Diffraction Based on Neural Network Algorithms

Fresnel diffraction, as a type of diffraction, presents significant mathematical complexity, making quantitative analysis challenging at the undergraduate level, and related experiments are relatively rare. Neural networks, which are algorithms that simulate the structure and function of human brain neurons, can be used to process complex data. In this paper, an irregular aperture is used as the object and a multi-aperture plate as the sampling plate in a lensless coherent optical setup. Neural network algorithms are employed to analyze the diffraction images, the morphology recognition of the object is achieved by calculating the inversion matrix. Additionally, detailed calculations and explanations are provided for the spacing of diffraction bright spots and the light intensity distribution. The designed Fresnel diffraction experiment in this paper is suitable for undergraduate teaching, helping students better understand and apply relevant theoretical knowledge.