The breakthrough of deep learning has started a technological revolution in various areas such as object identification, image/video recognition and semantic segmentation. Neural network, which is one of representative applications of deep learning, has been widely used and developed many efficient models. However, the edge implementation of neural network inference is restricted because of conflicts between the high computation and storage complexity and resource-limited hardware platforms in applications scenarios. In this paper, we research neural networks which are involved in the acceleration on FPGA-based platforms. The architecture of networks and characteristics of FPGA are analyzed, compared and summarized, as well as their influence on acceleration tasks. Based on the analysis, we generalize the acceleration strategies into five aspects—computing complexity, computing parallelism, data reuse, pruning and quantization. Then previous works on neural network acceleration are introduced following these topics. We summarize how to design a technical route for practical applications based on these strategies. Challenges in the path are discussed to provide guidance for future work.

In this Letter, we investigate a snapshot spectral-polarimetric-volumetric imaging (SSPVI) system using a single detector. Through compressed acquisition and reconstruction, SSPVI can achieve spectral imaging ( x , y , λ ) , polarization imaging ( x , y , ψ , χ ) , and light field imaging ( x , y , θ , φ ) simultaneously. The newly discovered performance is showcased by attaining the spectral-polarimetric-volumetric video and different laboratory accuracy experiments. These never-seen-before capacities of the camera open new prospects for many applications, such as biological analysis, object recognition, and remote sensing.

Aiming at addressing the vulnerability of the infrared imaging opto-mechanical system to stray radiation interference caused by an external laser, a new method with a broadband and high-absorptivity ceramic coating on the inner wall of the system is proposed to reduce the stray radiation intensity, which is helpful to improve the imaging quality of the detection image. Based on plasma electrolytic oxidation (PEO) technology, the preparation method and properties of a novel ceramic coating are studied, and a long-wave infrared imaging optical system with high-absorption ceramic coating is designed and developed. It is verified for the first time that the high-absorption ceramic coating can suppress the stray radiation in the infrared opto-mechanical system, and the effect of laser incident power and angle on the stray radiation is investigated. The experimental results show that the ceramic coating can achieve 95.4% high absorption in the wavelength range of 0.2–16 μm, which can obviously suppress the stray radiation in the opto-mechanical system of infrared imaging caused by laser. The calculation of the stray radiation suppression ratio shows that, compared with the case without coating, the stray radiation intensity in the system can be reduced by 70% using the ceramic coating.

The current measurement system of surface scattering rate applied in laser protective materials has the defects of low accuracy, discontinuous diagnosis region and narrow infrared measuring waveband. In order to make up for these shortcomings, a high-precision material-surface-scattering-rate measurement system based on a three-hole integrating sphere is proposed, which can realize the high-precision quantitative measurement on any region of coating surface from near-infrared to far-infrared band. Firstly, a new quantitative relationship between the luminous flux received by detector and the surface scattering rate of coating is obtained by modifying the existing integrating sphere scattering model. Secondly, a high-precision scattering characteristic measurement system based on a three-hole integrating sphere is designed and achieved. The influence of the main design parameters of the integrating sphere on the expected measuring accuracy of the system is investigated by using a TracePro simulation. Accordingly, the optimal design parameters of the system are given. Then, the main sources of the relative measurement uncertainty for the scattering rate are investigated experimentally, and four main relative uncertainty factors are evaluated quantitatively. Finally, according to the error propagation theory, the total experimental relative measurement uncertainty of the system is obtained, which is ±2.22% and 26–56% higher than the current measuring accuracy. The new coating-scattering-rate measurement system proposed in this paper can provide an effective experimental detection means for high-precision quantitative measurement and a performance evaluation for laser-protective-coating surface-scattering rate.

Aiming at the problem that the vertex detection method is difficult to deal with the high-efficiency detection of the large-scale spherical lenslet array, a contact off-axis measuring method is proposed and the measurement accuracy and detection efficiency are verified by experiments. Firstly, by analyzing the 3D model of the relationship between a spherical lens and probe of profilometer, the mathematical model of probe trajectory arc is established based on the off-axis trajectory characteristics of the spherical lenslet array. Then, the error mechanism under the off-axis condition is analyzed, and a mathematical algorithm is proposed to restore the structural parameters at the main optical axis of the lens by using the process design value of the unit lens aperture. Finally, a comparative experiment is carried out between the off-axis detection method and vertex detection method. The experimental results demonstrate that: Compared with the coaxial detection method, the relative errors of the measured lens curvature radius R₀ and the lens vector height f₀ under the off-axis detection method are 1.71% and 1.95%, respectively. Under the sampling measurement scale, the detection efficiency of the off-axis detection method is 41 times higher than that of vertex detection method.