Applied Mechanics and Materials Vol. 740

Abstract: The flower bud ratio is an important criteria and nodus in flower automatic grading realization, however using computer vision to detect flower bud still remains a challenge. This paper presents a flower bud detection model as a resolution: By using spectral residual method to calculate saliency map and using SURF feature-points, we can quickly and unambiguously obtain the SURF histogram, then construct the flower bud detection model with SVM. The experiments indicates that the flower bud detection model which provides an effective way for measuring bud ratio in flower automatic grading system can distinguish between bud and bloom with good results.

Abstract: For the needs of laser autocollimator in image acquisition and processing, designed the hardware circuit based on FPGA and DSP as the control core, detailed the system hardware composition. Image sensor initialization and configuration were elaborated, cached the image data by double buffering technique. By reading the cache data, pretreatment, spot extraction, calculating barycentric coordinates and data fitting, the high precision of angle measurement was obtained. The system design is reasonable, the hardware circuit is reliable, and the video capture image is clear. The functionality of image acquisition and processing is achieved.

Abstract: Aluminum alloy structure impact localization system by using fiber Bragg grating (FBG) sensors and impact localization algorithm were investigated. The impact localization method was proposed based on Extreme Learning Machine (ELM). Cross-correlation analysis method was used to extract the impact signal time difference. And ELM was used to realize impact localization. ELM model’s input was signal time difference and the output was the impact location. At last, FBG impact localization system was established. In aluminum alloy plate’s 500mm*500mm*2mm experiment area, the FBG impact localization system was used to identify the impact location. The experimental results showed that the impact location abscissa and ordinate localization errors were both less than 10mm. The research results provided an alternative method for the aluminum alloy material structure impact localization.

Abstract: Multi-object tracking has been a challenging topic in computer vision. A Simple and efficient moving multi-object tracking algorithm is proposed. A new tracking method combined with trajectory prediction and a sub-block matching is used to handle the objects occlusion. The experimental results show that the proposed algorithm has good performance.

Abstract: We consider a steady process of filtration of the incompressible high-viscosity fluid, following non-linear filtration law. Generalized statement of this problem is formulated in the form of operator equation with monotone operator in Banach space. To solve the operator equation, we suggest iterative method that does not require the inversion of the original operator. Each step of the iterative process can essentially be reduced to the solution of the boundary-value problem for the Laplace operator. This method was realized numerically. The numerical experiments made for the model problems confirmed the efficiency of the iterative method.

Abstract: There has been a tight correlation between the change of the cement rotary kiln carrent and operation conditions, such as the kiln thermal condition, the thickness of the crust, the kiln tyre collapse and so on. Considering the different fractal characteristics that cement rotary kiln current signal exhibit in different conditions, this paper has study on the generalized fractal dimension of kiln current signal under different working conditions as the characteristic parameter. The levels and trends of rotary kiln current and working condition will be identified by computing the maximum correlation coefficient according to the generalized fractal dimension of detected signal and of cement rotary kiln current under different working condition.

Abstract: Aiming at the problem that traditional median filter algorithm cannot process collected images quickly and efficiently, this paper adopts improved median filter and makes use of advantages, such as fast running speed, parallel running of inner program, to design an image preprocessing system with high real time ability and high flexibility. At last, compared with MATLAB median filter simulation figure and multilevel median filter, it has shown that using FPGA to process and improve median filter can not only conduct median filter to images successfully, but also has the ability of fast operation speed and low energy consumption.

Abstract: A detection method is given in this paper for the automatic surface solder printing quality detection during the automatic welding on PCB. The method of variable metric is used in the processing of the target positioning and the target recognition is used based on the machine vision. The method given here can greatly improve the processing speed in real. The processing of the rotating angle axis positioning theory is mentioned and the verification is done by experiment. This method has the advantage of high accuracy, small detection error and easy to operate.

Abstract: To study the problem of the identification of multiple ear tags of cattle occur data collision in the RFID network of pasture, the experiment use an improved anti-collision algorithm, it is combined bit locking with dynamic and analysis method, only colliding bits are processed by the method of binary, during transmission the data are dynamic, only send known information, which receive complementary bits of the known information, and then analyze the performance of the algorithm and simulation. The results show that compare with traditional binary algorithm, the improved algorithm from the operating frequency, transmission delay, data throughput have obvious advantages, the improved algorithm that could solve the problem of data collision cause by the cattle those wear ear tags get into radio frequency region, improve the efficiency of the tag identification.

Abstract: As to the problem of blind recognition for convolutional code, the existing recognitory methods’ principle and process are summarized, and they are simulated and analyzed with Matlab. Four methods are introduced for the recognition of generation polynomials, and the comparison of these methods shows that the Euclidean algorithm works better at low BER, while the method based Walsh-Hadamard transform has a better accuracy when BER is as high as 10^-2.