As for HEVC (High Efficiency Video Coding) standard decoder, high throughput rate and high
memory access volume of data serve as bottleneck problems, design of a HEVC-oriented high-efficiency sub-pixel
interpolation filtering VLSI (Very Large Scale Integration) architecture is put forward. Based on sub-pixel
interpolation algorithm in HEVC standard, an interpolation filtering VLSI architecture characterized by high
degree of parallelism and flow line is constructed. Through utilization of inversion symmetry of filter coefficients,
a reusable order-8 filter structure is designed in order to reduce area of filter hardware. On the basis of the
traditional single-input channel interpolator, a parallel-channel 8-input interpolator is designed to improve data
throughput. Experimental result indicates tha the design can meet decoding requirements of 1 920伊1 080@
30 f/ s video at a frequency of 34. 2 MHz. And this design can meet requirements of real-time transmission of
3 840伊2 160 @60 f/ s video.

 In order to optimize the performance of four-quadrant analog multiplier, to meet the application
requirements of modern analog signal processing circuit such as high-frequency and low-noise, a new low voltage
four-quadrant analog multiplier is proposed. The multiplier circuit uses the analog squarer based on MDDCC
(Modified Differential Difference Current Conveyor) as a basic module. The function of multiplying circuit is
confirmed by PSPICE ( Personal Simulation Program with Integrated Circuit Emphasis) simulations using
TSMC 0. 18 μm CMOS technology. The proposed circuit has a good linearity. The range of input voltage is
-0. 1 V ~ +0. 1 V, the -3 dB bandwidth of the proposed circuit is 451. 307 MHz and the output noise voltage is
less than 150 nV when the peak value of input voltage is 100 mV. By comparing with some available multipliers
in references, the proposed multiplier circuit has better linearity, higher cut-off frequency and reduces the output
noise voltage. The four-quadrant analog multiplier has good performance in the high frequency signal processing
system.

In order to realize the identification of panax ginseng, an intelligent automatic detection system is
developed. The hardware of the system mainly includes the driving circuit of the gas sensor array, signal
collecting and conditioning circuit, information processing circuit, air inlet part control circuit and external
interface circuit. The system uses PCA (Principal Component Analysis) and SVM (Support Vector Machine)
algorithm to realize the identification of ginseng category. The intelligent control of the temperature of the gas
chamber is realized by PID (Proportion Integral Differential) algorithm. And the stability and Accuracy of the
gas sensor is improved. The results showed that the system had high accuracy in the identification of Panax
ginseng. The red ginseng, white ginseng, American ginseng, heterophylla, Panax notoginseng and the
trichosanthin of similar shape and odor are used to test by the system. The experimental results show that the
system has a high accuracy in the identification of ginseng, and it is simple in operation and the cost is low, and
has good application prospect.

 In order to solve the problem of the low sensitivity of the traditional waveguide sensor based on
organic substrate, the MZI ( Mach-Zehnder Interferometer) waveguide sensing structure of high sensitivity
based on polymer materials is proposed, and two high sensitivity waveguide sensing structure with symmetric
MZI and asymmetric MZI is designed. The sensitivity of the optimized symmetric MZI waveguide sensor in the
interval 1. 34 ~ 1. 36, 1. 36 ~ 1. 38, 1. 38 ~ 1. 40 are 1 850. 76 dB/ RIU, 1 995. 09 dB/ RIU,
1 547. 30 dB/ RIU. The sensitivity of the optimizedasymmetric MZI waveguide sensor in the interval 1. 33 ~
1. 35,1. 35 ~1. 38,1. 38 ~1. 40 are 1 719. 23 dB/ RIU, 1 251. 68 dB/ RIU, 2 065. 22 dB/ RIU. For the two
sensing structure, the high sensing accuracy in 1. 33 ~ 1. 40 can be realized by the design of integrated
waveguide sensors.

 In order to solve the transmission inconvenience caused by cable data transmission in runoff sampling,
based on a wireless local area network (ZigBee) technology, a low-power sampling system for water and soil
erosion in field runoff is designed. It is used to automatically collect the surface runoff samples and to record the
sampling time and other parameters. TI company’s CC2530 microprocessor is used as the core controller which is
responsible for the wireless data transceiver and sampling device control. The system is composed of a wireless
handheld device and a plurality of runoff sampling devices. Wireless hand-held device and runoff sampling
device are composed of star ZigBee network for data transmission, the hand-held machine as the center node is
responsible for the management of the entire network and the control of each sampling device. The runoff
sampling device is a network node points separating runoff automatically according to the central node sampling
rules. The system uses 12 V lead acid battery as the power supply, and realizes the low power design of the
hardware and software. Field experiment shows that the system has the advantages of convenience, stability and
reliability.

 In traditional multimedia conference and teaching,presenter and teacher could only share message to
listener. In order to solve this single shared problem, we design a picture shared system based on embedded system,
Android system and iOS system. The system is divided into mobile terminal, Linux embedded system and server three
partitions. The mobile terminal compresses multimedia picture, and then uploads it to server with the form of stream.
The embedded system reverts the stream to pictures and shows on it’s projector screen. Real-time refresh shared
images, and new shared images overwrite previously shared images. Server saves all the pictures into database, so all
the mobile terminals could review or save them. The results show that real-time sharing of different pictures can be
realized on the embedded platform, and the real-time sharing of multimedia file information is solved.

In order to reduce the LDPC (Low-Density Parity-Cheek) complexity of decoding algorithm, studied
the LLR BP (Log-Likelihood Ratio Belief Propagation algorithm) and its simplified decoding algorithm, and an
improved algorithm based on NMS (Normalized Min Sum) algorithm is proposed; studied the Gamma-Gamma
distribution model and Gamma-Gamma distribution as the channel model are studied. The performance of LLR-
BP and its simplified algorithm under different turbulence intensities are compared with Matlab, the simulation
results show that both in the AWGN (Additive White Gaussian Noise) or different turbulence intensity Gamma-
Gamma channel, the improved algorithm has better performance in decoding algorithms in particular, the
performance advantages of the improved algorithm in strong turbulence is more obvious. The simulation results
show that the increase of the code length can improve the performance of the BER (Bit Error Rate).

 In order to solve the position control and anti swing control in the horizontal movement of overhead
crane, we used Lagrange equation to establish its nonlinear mathematical model, and designed PD (Proportional
Differential) controller and ADRC (Auto Disturbances Rejection Controller). The parameters were optimized by
artificial fish swarm algorithm, overcoming the complexity of trial and error method. The controller was tested by
simulation experiment, the simulation results showed that this controller system could effectively decreases
winging angle of the load, and had strong robustness to system parameter svariation and swinging angle
disturbance.

 In order to solve the problem of the synchronization of the biological mathematical model of muscular
blood vessel, the tracking control method is used for the first time,constructing a proper chaotic synchronization
controller by the Lyapunov stability theory, making the drive-response system achieve synchronization effect in
the global scope,and the simulation experiment was carried out for the drive-response system. Experimental
results show that the synchronization error of the system gradually converge to zero, and it can make the blood
vessels under the spasticity and normal blood vessels achieve synchronous motion by the control system, and the
effectiveness of the control method is proved, providing a theoretical basis for the treatment for the disease of
vasospasm.

 In order to solve the shortcoming of fewer extracted line feature and lower recognition rates of
pathologic palmprint recognition algorithm, we proposed a recognition algorithm of pathologic palmprint based on
spectral fusion in non-subsampled shearlet domain. Firstly, the best spectral fused combination of multispectral
palmprint is selected. And it is decomposed to the multi-directions, multi-scales in non-subsampled shearlet
domain. Next, according to all levels characteristics of sub-bands images which had been decomposed, a new
fusion rule is designed to fuse the corresponding coefficient matrices. The fine lines feature of palmprint can be
obtained by the inverse transformation of the NSST(Nonsubsample Shearlet Transform) and the process of
mathematical morphology. Then the satisfactory closed circuits are searched by degree feature of pixels. Finally,
we proposed a method of combining the shape descriptors based on rectangle degree and eccentricity to recognize
closed pathologic palmprint. Experimental results show that this algorithm can extract rich feature of the
palmprint line, and can recognize six different types of closed pathologic palmprints accurately and the
recognition rate is more than 90%.

In order to improve the virtual-real registration efficiency of augmented reality system for medical
auxiliary operation, a regular hexahedron three-dimensional marker based on hamming code put forward as the
characteristic pattern was designed. The system detected and identified the marker, and POSIT(Pose from
Orthography and Scaling with Iterations)algorithm was used to obtain the camera position for the characteristics of
the three-dimensional marker and the size of the marker was optimized by experiment. The virtual skull model
after CT(Computed Tomography) reconstruction was real-time displayed in the real scene using the transfer
matrix obtained from POSIT. Experiments proved the marker’s effectiveness.

 To prevent the influence of flight safety caused by damage of key components of hypersonic aircraft wing
effectively, the damage evolution of key components of hypersonic aircraft wing and the effects of the flight
dynamics to the damage are studied. By establishing the damage model of the aircraft wing based on the material
parameters and computing the loading stress of the wing’s root, the effects of flight height, velocity, angle of attack
and aileron deflection angle to the damage characteristic are analyzed to determine the crucial variable which affects
the damage. Simulation results show that the angle of attack plays a major role in the damage evolution compared
with other variables. Therefore, in hypersonic flight, the angle of attack should be limited as far as possible to
guarantee the safety of the vehicle and extend its life span. The achieved results lays a solid foundation for the
damage-mitigating controller design, which also provides an important reference to enhance the structural reliability
and extend the life span for hypersonic aircraft.

 In order to meet the requirements of electric vehicle battery thermal management system which is
purposed to reduce the maximum temperature of the battery pack and the temperature difference between single
cells, to overcome the shortcomings of existing air cooling system and series loop water cooling system, to reduce
the maximum temperature and the temperature difference between single cells, a water cooling system with
parallel loop is proposed. Based on the analysis of the heat generation mechanism of lithium ion battery, the
temperature model of the battery was established, and the battery water cooling system with parallel loop was
built in AMESim (Advanced Modeling Environment for Performing Simulation) software. The heat dissipation
performance was compared with the series loop water cooling system via simulation experiments. Good results
have been achieved, the temperature difference is effectively reduced on the basis of maintaining the maximum
temperature of the battery pack, which laid the foundation for the further study of the control algorithm for the
parallel loop water cooling system.

 A QoS link selection algorithm based on multi-constraint condition feedback is proposed for the link
selection in the ubiquitous positioning system of Internet of Things to satisfy the requirement of multiple
restrictions. By using the characteristic of wireless communication broadcasting, the problem of the multi-
constraint service quality routing is transformed into the problem of minimum cost in satisfying the link length,
bandwidth, delay, delay jitter, packet loss rate, and cost and so on. In this algorithm, the multi-hop
autonomous link model of Internet of Things is established and the target energy function is defined, and the
Hopfield neural network model is used to transform multiple constraint conditions into a polynomial extremal
problem. The algorithm avoids the problem of selecting a large number of parameters without changing the system
link selection strategy. It realizes the automatic selection of link and satisfies the requirement of real-time
communication.

 In order to solve the problem of low precision in traditional text localization algorithm, a robust text
localization method based on visual saliency and edge density is proposed from the perspective of natural scene
text properties. Firstly, spectral residual theory is used to extract the saliency region of the image, then search
for the regions of large edge density in the saliency regions and construct candidate connected regions, using a
small amount of prior information to filter out non text area. The experimental results on the standard data set
show that this method can obtain 70% comprehensive detection rate compared to the method of text region
detection using edge feature.

 In order to solve the problem of slow convergent speed and poor local search ability of CS(Cuckoo
Search) algorithm, a self-adaptive step is introduced in CS, which balances the global and local search
ability and improves the convergent speed. Experiments have been conducted on functions. The results
indicate the good performance of the new algorithm. Initial weights and threshold values are optimized by the
modified CS. Compared with the conventional BP ( Back Propagation) algorithm, the condition of slow
training speed and premature result is released. The new neural network is used in pump units蒺 fault
diagnosis. The result shows that CS-BP enhances the convergence rate and robust of neural network, and the
accuracy of the fault diagnosis.

 The traditional programming method is rigid and inflexible to implement the programmers’thoughts,
and lacks reasoning ability. And it changes greatly when code modification is needed, the expandability is poor.
A new programming language action language C+ which is a declarative programming method is made use of
to represent and solve issues. Taking the pouring problem as an example, the main rules and query are
represented by C+, then call the cplus2asp to reason automatically. The experiment certifies that the syntax of
C+ language is simple and the solving process has the ability of automatic reasoning, it can add or delete
statements arbitrarily without affecting the execution of other statements, having high expandability.

 Aiming at the traditional support vector regression machine on the mechanism can’t solute dynamic
time-varying signal pattern classification, proposes a process support vector regression time series prediction
model, and the vortex search algorithm for support vector machine parameter optimization. Using air quality data
set of UCI(University of California Irvine)machine learning repository and belgium solar influences data analysis
center sunspot activities data for simulation. The simulation results show that the prediction results of the
prediction model are better than the particle swarm optimization process support vector regression and support
vector regression, the time series prediction model has well predictive ability.

 Moving target tracking in video is an important research direction in the field of computer vision. In
order to detect and track the moving objects more accurately and quickly, we propose an improved CAMSHIFT
(Continuously Adaptive Mean-SHIFT) algorithm based on background subtraction and frame difference algorithm.
First of all, the target area is determined by the background difference method and the frame difference method,
then combining with the CAMSHIFT iterative algorithm the target tracking is realized. Experiments show that this
method has changed the traditional CAMSHIFT algorithm which needs manually selecting the target, has overcome
the limitations of diverging, and has improved the accuracy and stability of tracking.

 In order to express the high frequency minutia information of the image in detail, and improve the
quality of reconstructed image, a super resolution reconstruction algorithm is applied based on multi dictionary
L1/2 regularization. In the dictionary training phase, in order to effectively extract the information of feature
detail of edge and texture of low resolution image, the modified first or second order method is used to extract
feature for low resolution image. In the stage of image reconstruction, because of the problems that the solution
based on L1 regular model is usually not sparse enough and the quality of the reconstructed image needs to be
further improved, L1/2 norm is employed to substitute L1 norm to establish the super resolution reconstruction
model. The experiment shows that the present algorithm compared with the existing algorithms can better express
the section information of the image details and improve the quality of image reconstruction.