detection methods based on deep learning are the current research focus of computer vision. However, when detecting small objects, existing detectors often suffer from missing detection. Every pixel of hyperspectral images contain the spectral information of small object materials. Therefore, they can provide additional support for improving the detection performance on small objects. However, the adjacent bands of hyperspectral images are highly correlated. It is thus necessary to select representative bands to reduce the computational redundancy. In response, this paper proposed a hyperspectral small object detection model, which used the radial basis activation function (RBAF) to carry out spectral screening and object detection. Specifically, in view of the band redundancy of hyperspectral images, an attention mechanism based on the RBAF was designed for spectral screening. As for the high texture fuzziness and low distinguishability against the background of small objects, the resolution of input images was reconstructed first. Then, a radial basis object output network (RBOON) based on the RBAF was constructed to enhance object classification. For model simplification, spectrum screening and resolution reconstruction were integrated into an attention-based resolution reconstruction network (ABRRN). With the combination of the ABRRN and RBOON, the detection model can screen the specific spectrum and suppress false alarms and thus improve the accuracy of small object detection. Hyperspectral small object detection experiments show that the proposed method improves the two detection accuracy criteria, namely AP50 and AP50:95, by 5.4% and 0.2%, respectively, which means it is better than other methods.

ing at the technical problems of existing unmanned aerial vehicle (UAV)-borne hyperspectral imagers, a compact visible near infrared imaging spectral system is designed. First, an imaging spectrometer and a plane array camera are integrated to design the common optical path. Then, the high frame frequency plane array image is used to invert the position and attitude parameters of the camera. Finally, the high-precision spatial information correction is carried out for the synchronously obtained push and sweep hyperspectral image. The system has a working range is 400--1000 nm, a field of view is 43.6° in the width direction, a field of view is 20.0° in the flight direction, a focal length is 13 mm, and a spectral resolution is better than 2.5 nm. ZEMAX software is used to optimize the design and analysis of the system, and prism-grating prism (PGP) design is used for the spectrometer, which has the characteristics of light weight, low cost, and high resource utilization.

ing at the quality of the light beam output by the silicon-based optical phased array chip affected by the phase noise during the waveguide etching and bonding process, a phase controller is developed based on the stochastic parallel gradient descent algorithm with a high-speed single-point photodetector as the performance evaluation function collector. Construct a one-dimensional 64-element silicon-based optical phased array chip beam optimization principle experimental system, and realize the fast phase noise compensation of silicon-based optical phased array chip. Then, we study the influence of the matching relationship between the performance of the phase controller and the response characteristics of the silicon-based optical phased array chip on the beam optimization effect, the influence of the initial light field intensity on the convergence time of the stochastic parallel gradient descent algorithm, and the influence of the ambient temperature on the phase noise compensation effect. The experimental results show that the time for the system to complete the single-angle beam optimization is 0.26 s, and the optimized beam indicators are consistent with the device design.