In the natural world, insect compound eyes (CEs) offer significant advantages in visual information acquisition [[1], [2], [3]]. Typically, CEs consist of numerous independent photosensitive units, known as ommatidia, which work in concert to provide a wide field of view (FOV) and capture substantial spatial information. This structure enables insects to exhibit heightened motion detection and depth perception, allowing them to quickly and accurately determine the position of objects in various directions [[4], [5], [6]]. These capabilities are critical for insects in locating food and avoiding predators. Inspired by the sophisticated structure of natural CEs, researchers are actively developing bionic artificial CE imaging systems, with promising potential applications in areas such as micro-robot vision, medical endoscopy, and virtual reality [[7], [8], [9], [10], [11], [12], [13], [14], [15]].

Bionic artificial CE lenses offer several advantages over conventional lenses, including a wide FOV, low aberration, compact size, and ease of integration. Considerable effort has been made to develop CE imaging systems that match or surpass the performance of natural CEs. In 2005, a planar-CE imaging system consisting of 100 cameras was developed, resulting in a 57° FOV [16]. However, this macroscopic camera array is unsuitable for miniaturization and integration into compact imaging systems. In contrast, micro-CE lenses have garnered more attention. Duparré et al. presented a micro artificial apposition CE system with a 21° FOV and a system thickness of only 320 μm, sufficient for many micro sensor and imaging applications [17]. Dunkel et al. further developed a planar CE by combining a refractive freeform array with a conventional planar microlens array, extending the FOV to 70° [18]. These approaches have successfully broadened the FOV while enabling low-aberration imaging. However, the planar arrangement significantly restricts the potential FOV of CE systems. Micro-curved CE lenses, where sub-lenses are arranged on a curved surface, allow the collection of incident light from multiple directions, thus achieving a wider FOV [19]. However, fabricating high-quality curved CEs poses significant manufacturing challenges, with traditional techniques struggling to meet the required precision and 3D molding capabilities. In 2014, Wu et al. utilized femtosecond laser 3D printing to create a curved CE lens with a 90° FOV [20]. Subsequently, in 2016, Chen et al. developed a bionic artificial CE using single-pulse femtosecond laser wet etching and hot embossing, achieving a large 140° FOV with low aberration [21]. These studies produced high-quality micro-curved CE lenses and demonstrated large FOV imaging capabilities. However, the simple spherical or aspherical surfaces of traditional CE sub-lenses limit the imaging depth of field (DOF). To address this, Ren et al. developed a liquid lens capable of adjusting focal length by applying voltage to the droplet, effectively extending the imaging DOF [22]. Similarly, in 2011, Zhang et al. introduced an integrated adjustable microlens that enables focal length adjustments between 3 mm and 15 mm via electrothermal actuation, significantly improving the imaging DOF [23]. Then, Ma et al. developed a curved CE lens with adjustable FOV and focal length, driven by environmental pH changes based on bovine serum albumin (BSA) [24]. Recently, Cao et al. achieved variable focal length imaging by integrating a deformable microlens array and a microfluidic chamber [25]. While these advancements have greatly enhanced imaging DOF through the development of adjustable micro-CE lenses and optimized lens designs, their implementation remains constrained by reliance on mechanical drives or environmental changes. These limitations result in inflexible imaging, slow response times, and integration challenges.

On the other hand, a complete imaging system consists of two key components: the imaging lens and detection system. Researchers have attempted to integrate CEs with detectors to create functional imaging systems, but they face a significant challenge for defocusing. Specifically, when imaging an object through sub-lenses on a spherical surface, the resulting image is projected onto a curved focal plane, which a flat detection system cannot capture in a single location. This issue has motivated considerable efforts to find effective solutions. One approach to solving the defocus problem is the incorporation of an image transfer function between the lens and the image receiver. For instance, Zheng and Dai et al. introduced a conical light guide device with image transfer capabilities, which they placed between the curved CE lens and the detector. This design allowed them to project CE images onto a flat surface [26,27]. Another approach involves using multilayer lens combinations to project images onto a flat plane. Zhang et al. developed a three-layer 3D artificial CE to distributed incoming light across a planar detector [28]. Similarly, Toulouse et al. addressed the defocus issue by creating a lens combination with a free-form surface, employing femtosecond laser 3D printing technology to obtain the required precision [29]. However, both methods face challenges in terms of precise manufacture, alignment and complex assembly processes. Additionally, Sun et al. tackled the defocus issue by designing the focal length of each sub-lens according to its position within distinct layers, which improved performance but imposed higher demands on lens design and fabrication [30]. In our previous work, we proposed a logarithmic curved CE lens that addressed the optoelectronic integration of curved microlenses and microdetectors by extending the focal depth. While this design, featuring uniformly curved sub-lenses, simplified manufacturing and design complexity, the increased focal depth resulted in a reduction of imaging quality [31].

In summary, significant efforts have been made to develop miniature CE imaging systems with a wide FOV, low aberration, and large DOF, resulting in several breakthrough research achievements. However, most reported studies have focused on enhancing individual performance metrics, and the development of a fully integrated miniature CE imaging system with no defocus, large DOF, and wide FOV remains a challenging task. Here, inspired by the CEs of trilobites, which feature two distinct photosensitive components [32], we propose a micro-CE imaging system that consists of customized bifocal microlenses arranged on a spherical curved surface. The focal lengths of the sub-lenses at different positions on the curved surface are meticulously designed to project targets from various directions onto a single plane. Additionally, each sub-lens possesses two distinct focuses, enabling the imaging of both near and distant objects, thus significantly extending the DOF. This specially designed micro-curved CE was fabricated using femtosecond laser 3D printing technology, ensuring high surface precision [[33], [34], [35], [36]]. Compared with the traditional compound lens, the bifocal curved compound lens proposed in this paper has the advantages of small size, large FOV, large DOF imaging characteristics, and easy integration with miniature image detectors. As a proof of concept, a micro-curved CE with a diameter of 300 μm was successfully created and evaluated. Experimental results confirm that the CE lens can simultaneously capture images with a 90° FOV for objects ranging from near (<200 μm) to far (>10 cm) distances. Our work provides an exemplary approach to achieving large DOF and wide FOV imaging without mechanical adjustments or external environmental changes, offering promising applications in high-performance micro-imaging systems, such as medical endoscopes and microrobot vision.