Species discrimination of insects is an important aspect of ecology and biodiversity research. The traditional methods based on human visual experience and biochemical analysis cannot strike a balance between accuracy and timeliness. Morphological identification using computer vision and machine learning is expected to solve this problem, but image features have poor accuracy for very similar species and usually require complicated networks that are unfriendly to portable edge devices. In this work, we propose a fast and accurate species discrimination method of similar insects using hyperspectral features and lightweight machine learning algorithm. Feature regions selection, feature spectra selection and model quantification are used for the optimization of discriminating network. The experimental results of six similar butterfly species in the genus of Graphium show that, compared with morphological recognition with machine vision, our work achieves a higher accuracy of 92.36 ± 3.04% and a shorter inference time of 0.6 ms, with the tiny-size convolutional neural network deployed on a neural network chip. This study provides a rapid and high-accuracy species discrimination method for insects with high appearance similarity and paves the way for field discriminations using intelligent micro-spectrometer based on on-chip microstructure and artificial intelligence chip.

Infrared video imaging is an cornerstone technology for environmental perception, particularly in drone-based remote sensing applications such as disaster assessment and infrastructure inspection. Conventional systems, however, rely on bulky optical architectures that limit deployment on lightweight aerial platforms. Computational imaging offers a promising alternative by integrating optical encoding with algorithmic reconstruction, enabling compact hardware while maintaining imaging performance comparable to sophisticated multi-lens systems. Nonetheless, achieving real-time video-rate computational image restoration on resource-constrained unmanned aerial vehicles (UAVs) remains a critical challenge. To address this, we propose Mobile Wavelet Restoration-Net (MWR-Net), a lightweight deep learning framework tailored for real-time infrared image restoration. Built on a MobileNetV4 backbone, MWR-Net leverages depthwise separable convolutions and an optimized downsampling scheme to minimize parameters and computational overhead. A novel wavelet-domain loss enhances high-frequency detail recovery, while the modulation transfer function (MTF) is adopted as an optics-aware evaluation metric. With only 666.37 K parameters and 6.17 G MACs, MWR-Net achieves a PSNR of 37.10 dB and an SSIM of 0.964 on a custom dataset, outperforming a pruned U-Net baseline. Deployed on an RK3588 chip, it runs at 42 FPS. These results demonstrate MWR-Net’s potential as an efficient and practical solution for UAV-based infrared sensing applications.

Diffractive lens (DLs) computational imaging promises lightweight infrared imaging. However, it is still challenging to accurately obtain the degradation of DLs computational imaging systems to achieve high-quality imaging. Degradation estimation methods based on image-to-image are the most feasible way, but existing methods face the problem of low accuracy for severe degradation estimation. In this paper, we proposed a severe degradation estimation framework based on KernelNet, and realized the improvement of the accuracy of severe degenerate estimation by more than 1.13 dB. Furthermore, DyUnet, an image restoration network, was proposed, which is based on dynamic convolution. The results of the mid-wave diffractive lens infrared computational imaging system indicated that our method achieved high-quality imaging results comparable to the conventional one, while the number of optical elements was reduced from 7 to 2, and the weight was reduced by 50%.

Platycodonis radix is a commonly used traditional Chinese medicine (TCM) material. Its bioactive compounds and medicinal value are closely related to its geographical origin. The internal components of Platycodonis radix from different origins are different due to the influence of environmental factors such as soil and climate. These differences can affect the medicinal value. Therefore, accurate identification of Platycodonis radix origin is crucial for drug safety and scientific research. Traditional methods of identification of TCM materials, such as morphological identification and physicochemical analysis, cannot meet the efficiency requirements. Although emerging technologies such as computer vision and spectroscopy can achieve rapid detection, their accuracy in identifying the origin of Platycodonis radix is limited when relying solely on RGB images or spectral features. To solve this problem, we aim to develop a rapid, non-destructive, and accurate method for origin identification of Platycodonis radix using hyperspectral imaging (HSI) combined with deep learning. We captured hyperspectral images of Platycodonis radix slices in 400–1000 nm range, and proposed a deep learning classification model based on these images. Our model uses one-dimensional (1D) convolution kernels to extract spectral features and two-dimensional (2D) convolution kernels to extract spatial features, fully utilizing the hyperspectral data. The average accuracy has reached 96.2%, significantly better than that of 49.0% based on RGB images and 81.8% based on spectral features in 400–1000 nm range. Furthermore, based on hyperspectral images, our model’s accuracy is 14.6%, 8.4%, and 9.6% higher than the variants of VGG, ResNet, and GoogLeNet, respectively. These results not only demonstrate the advantages of HSI in identifying the origin of Platycodonis radix, but also demonstrate the advantages of combining 1D convolution and 2D convolution in hyperspectral image classification.

Encoded computational hyperspectral cameras, propelled by advances in compressed sensing theory, making both miniaturization and real-time hyperspectral imaging feasible. Spectral-encoded or spatial-encoded hyperspectral imaging strategy have limited numbers of design parameters in optical components, leading to severe ill-posedness in hyperspectral images reconstruction, which constrain overall imaging quality. However, spatial-spectral-encoded hyperspectral imaging strategy which simultaneously performs spatial and spectral encoding entailing more powerful modulation, alleviating ill-posed problems and improving the quality of hyperspectral images. In this paper, we present a co-modulation framework based on diffractive optical element (DOE) and Superposition Fabry–Perot (SFP) filter array for computational hyperspectral camera that integrates these two components with a transformer-based reconstruction network through end-to-end learning. The learned DOE and SFP filter encode the hyperspectral datacube on the sensor via phase and amplitude modulation, and the transformer-based network accurately reconstructs the images from sensor measurements. We conduct extensive simulations to analyze and validate the relatively contributions of the DOE, SFP filter, and transformer-based reconstruction algorithm to the significantly improved performance of hyperspectral image reconstruction across various ablation study models. We further investigate and identify the $\mathbf {4\times 4}$ SFP filter unit configuration as the most effective design for achieving a balance between spectral fidelity and spatial resolution. Our results show that the proposed system outperforms state-of-the-art methods in hyperspectral images reconstruction quality, excelling in both spatial and spectral detail recovery, and maintaining good performance against realistic noise levels.