Concerning the poor performance of the Single-Shot Refinement Neural Network for Object Detection (RefineDet) of the object detection network when detecting small sample classes in inter-class imbalanced datasets, a Subsection Weighted Loss (SWLoss) function was proposed. Firstly, the inverse of the number of samples from different classes in each training batch was used as the heuristic inter-class sample balance factor to weight the different classes in the classification loss, thus strengthening the concern on the small sample class learning. After that, a multi-task balancing factor was introduced to weight classification loss and regression loss to reduce the difference between the learning rates of two tasks. At last, experiments were conducted on Pascal VOC2007 dataset and dot-matrix character dataset with large differences in the number of target class samples. The results demonstrate that compared to the original RefineDet, the SWLoss-based RefineDet clearly improves the detection precision of small sample classes, and has the mean Average Precision (mAP) on the two datasets increased by 1.01 and 9.86 percentage points, respectively; and compared to the RefineDet based on loss balance function and weighted pairwise loss, the SWLoss-based RefineDet has the mAP on the two datasets increased by 0.68, 4.73 and 0.49, 1.48 percentage points, respectively.

Spiking neuron is a novel artificial neuron model. The purpose of its supervised learning is to stimulate the neuron by learning to generate a series of spike sequences for expressing specific information through precise time coding, so it is called spike sequence learning. Because the spike sequence learning for single neuron has the characteristics of significant application value, various theoretical foundations and many influential factors, the existing spike sequence learning methods were reviewed and contrasted. Firstly, the basic concepts of spiking neuron models and spike sequence learning were introduced. Then, the typical learning methods of spike sequence learning were introduced in detail, the theoretical basis and synaptic weight adjustment way of each method were pointed out. Finally, the performance of these learning methods was compared through experiments, the characteristics of each method was systematically summarized, the current research situation of spike sequence learning was discussed, and the future direction of development was pointed out. The research results are helpful for the comprehensive application of spike sequence learning methods.

In view of the problems that upper limit of radiated peak power is low for continuous wave laser radar, which limits the maximum measurement range in the application of range and speed measurement, a waveform of modulated signal based on Golomb series was proposed, and the feasibility of simultaneously measuring target's range and speed in road environments by the method was studied. Firstly, the problem of low transmitted signal peak power that exists in continuous wave modulating method was analyzed by using a quasi-continuous, i.e., Pseudo random Noise (PN) code modulation as an example. Characteristics of Golomb series were discussed, and a modulation method based on Golomb series was proposed for raising the peak power of transmitted pulse. Then, a method for analyzing spectrum of Doppler signal modulated by Golomb series was discussed, as well as a data accumulation method for locating signal delay time, such that range and speed could be measured simultaneously. Finally, within the range of Doppler frequency that is generated by moving road targets in road environment, simulations were performed to verify the correctness of the proposed method. The experimental results show that Fast Fourier Transform (FFT) can be used for obtaining the frequency of Doppler signal even when the sampling frequency provided by the pulse series is much lower than the Nyquist frequency, thus largely increasing the peak power of single pulse under the condition that average transmission power keeps unchanged. Furthermore, data accumulation method can be used for locating laser pulse flight time by exploiting the non-equal interval property of Golomb series, ensuring both target range measurement and speed measurement from the same signal.

For dealing with the problem of low efficiency and high maintenance cost when multi-point breakdown or change occurs in Wavelength Division Multiplexing (WDM) network with high speed and large capacity, the component of Reconfigurable Optical Add/Drop Multiplexer (ROADM) was used to construct a flexible network. Firstly, the 5-node network configuration model was provided. Then, the relation between loss and transmission length was investigated when optical network was composed of ROADM under dynamic conditions. The design flow of network transmission length was proposed. Next, a 5-node bi-directional fiber ring experiment network was constructed, and the optical loss characteristics were measured. Finally, based on the analysis of experiment data, it shows that the computed optical loss value and the measured loss value are approximately equal (0.8 dB difference). Thus, the feasibility of the design is verified, which assures the reliable transmission between nodes.

Concerning the key problems of selecting threshold function in wavelet threshold denoising, in order to address the discontinuity of conventional threshold function and large deviation existing in the estimated wavelet coefficients, a continuous adaptive threshold function in the whole wavelet domain was proposed. It fully considered the characteristics of different sub-band coefficients in different scales, and set the energy of sub-band coefficients in different scales as threshold function's initial weights. Optimal weights were iteratively solved by using interval advanced-retreat method and golden section method, so as to adaptively improve approximation level between estimated and decomposed wavelet coefficients. The experimental results show that the proposed method can both efficiently reduce noise and simultaneously preserve the edges and details of image, also achieve higher Peak Signal-to-Noise Ratio (PSNR) under different noise standard deviations.