Holographic Laser Processing: Rapid Manufacturing and Image Reconstruction of Artificial Biomimetic Compound Eyes

Introduction
In recent years, inspired by insect compound eyes, artificial biomimetic compound eyes have shown great advantages in overcoming the limitations of existing imaging devices such as large, bulky, and heavy, and improving the performance of medical endoscopy, panoramic imaging, micro navigation, and robot vision due to their unique optical imaging solutions such as small size, distortion free imaging, wide field of view, and highly sensitive motion tracking capabilities, attracting increasing attention.

The existing artificial biomimetic compound eye manufacturing mainly faces two challenges: firstly, the existing process is relatively complex, with limited manufacturing efficiency and difficulty in meeting commercial standards; The second issue is that the surface characteristics of the artificial biomimetic compound eye do not match those of commercial planar imaging sensors, making it difficult to integrate the two.

In view of this, Professor Chen Qidai from Jilin University proposed a wet assisted holographic laser processing method, which greatly improves the processing efficiency of artificial biomimetic compound eyes by customizing the preparation of biomimetic compound eyes and large-scale transcription. At the same time, combined with artificial intelligence methods, it solves the problem of mismatch between artificial biomimetic compound eyes and planar imaging sensors from an algorithmic perspective.

This research achievement is based on the "Holographic laser fabric of 3D artistic compound μ- Eyes ", published online in Light: Advanced Manufacturing.

Figure 1: Flow chart of wet assisted femtosecond laser parallel manufacturing of artificial biomimetic compound eyes

In the experiment, femtosecond laser without spatial light field modulator (SLM) modulation was first used to expose the surface of quartz substrate, and a compound eye main lens was formed using wet etching method; Utilizing SLM to split femtosecond laser beams and combining wet etching to achieve parallel processing of multiple small eyes in the compound eye; The large-scale production of compound eye microlens arrays can be achieved through PDMS micro/nanostructure transcription technology. The compound eye microlens array prepared by this method has the characteristics of high resolution and wide field of view. To overcome the difficulty of integrating artificial biomimetic compound eyes with planar cameras, high-quality image reconstruction was achieved through the use of generative adversarial networks (GAN), laying the foundation for future device integration.

Figure 2: Sample image of large-scale manufacturing of artificial biomimetic compound eyes
The complex optical devices manufactured by holographic laser processing technology are scalable. To address the complexity and time-consuming nature of the process, Figure 2 shows the large-scale production of polydimethylsiloxane (PDMS) soft micro optical components using quartz glass based compound eye microlenses as hard templates. During this process, the micro optical components still maintained high surface quality (as shown in SEM image 2a and 3D depth of field image 2b).

Figure 3: Image reconstruction based on generative adversarial network (GAN) deep learning algorithm
The curved contour gives the compound eye a wide field of view, but it also limits its focus position and can only be positioned on a curved focal plane. For a true biological compound eye, there is a fiber optic cable that can receive light and direct it into the retina. However, it is difficult to be compatible with current sensor programs, and it is also difficult to integrate optical components and detectors on the chip. In theory, the parameters of each small lens, including height, curvature, and focal length, should be redesigned, but it is difficult to determine the plane based on its position on the curve contour. To this end, we propose a deep learning algorithm based on generative adversarial networks (GAN) for image processing. In this study, we utilized two neural networks to maximize the generation ability of the discriminator and minimize its loss function, while the discriminator was trained to maximize its loss function. As shown in Figure 3a, after training, the neural network can use the image shown in Figure 3c to perform image restoration on all eyes. Image restoration is independent of incident wavelength, material refractive index, or single lens thickness. With this technology, compound eye imaging can preserve a large field of view and significantly improve image quality, making it suitable for a wider range of application scenarios (Figure 3b).

This study proposes an efficient femtosecond holographic laser method for preparing artificial biomimetic compound eyes, introducing artificial intelligence methods for reverse image reconstruction, solving the pain point problem of low manufacturing efficiency of artificial biomimetic compound eyes, and laying the foundation for future matching and integration of artificial biomimetic compound eyes with planar imaging sensors.

Paper Information
Lei Wang, Wei Gong, Xiao-Wen Cao, Yan-Hao Yu, Saulius Juodkazis, Qi-Dai Chen. Holographic laser fabrication of 3D artificial compound μ-eyes[J]. Light: Advanced Manufacturing 4, 26(2023).

https://doi.org/10.37188/lam.2023.026

Source: Advanced Manufacturing