Applied Mechanics and Materials Vols. 701-702

Abstract: This paper mainly performs Cascade AdaBoost algorithm based on multi-feature to detect the images of Eurydema dominulus, which will cause harm to crucifer. Firstly, the mixing of HAAR features and LBP features is adopted instead of the single-feature of traditional model, which makes description of images more comprehensively from the angle of the gradient and texture. And then use the best features selected by Gentle AdaBoost algorithm to compose the weak classifier and the strong classifier. And the cascade detector is composed of the trained classifiers of each layer according to a certain screening rate. Experimental results show that the method of detection has the probability of dis-detecting and leak-detecting, but it still has a certain reference value in the field of agricultural plant diseases and insect pests detection for its good robustness.

Abstract: Camera calibration is the most important stage of machine vision measurement. The principle and method of camera calibration for binocular stereo vision system are introduced and the left and right CCD are respectively calibrated by using the prepared calibration target and the MATLAB program. Then internal and external camera parameters are obtained by the calibration experiments. The experimental results show that the calibration results have high precision.

Abstract: This paper has a brief introduction to a algorithm that can automatically identify the defective cutting region when automatic paper cutter does quality analysis. The algorithm analyzed the characteristics of processed shear mark image,detected defects,found a critical point of defective part ,then a unique regional recognition algorithm was developed based on the basis of the general processing algorithm. The algorithms can pinpoint the region that cutting defects exist according to the coordinates of critical point accurately, improve the efficiency of quality analysis, and also has stated that image processing is available in the regional recognition of automatic paper cutter.

Abstract: Super-resolution image restoration technique can certainly increase the resolution of the obtained images without altering the system hardware conditions, which has become a research hotspot in the fields of remote sensing, military surveillance, public safety and medical imaging. The spatial domain algorithm has been studied widely and deeply, but it still has shortcomings and limitations. This paper studies the image degradation mechanism, establishes a precise image degradation process model, proposes an improved hybrid MAP-POCS restoration algorithm with using the Huber-Markov random field model as the priori probability model, which adds the convex sets constraints to the MAP estimation process and uses the peak signal to noise ratio (PSNR) to evaluate the recovery image quality. The simulation results show that the improved hybrid algorithm combines the only solution solving feature and noise reduction ability of the MAP method and strong prior knowledge utilized feature and flexibility of the POCS method, effectively ensures the convergence stability of the restoration and maintains the edge details of the restored image, enhances the effect of the super-resolution restoration.

Abstract: Since the infrared image has the problem of poor contrast, edge blur, low noise ratio, noisy and other shortcomings, this paper presents an improved threshold denoising algorithm based on curvelet transform, using a combination of hard and soft threshold methods to form a new threshold based functions. Compare to the conventional hard threshold, soft threshold, de-noising based on wavelet transform method, the proposed denoising method can obtain better PSNR and visual characteristics, retain more details

Abstract: As the fundamental work of Chinese information processing, Chinese word segmentation has achieved great progress since its birth. This paper reviews the research status of the CWS, discusses the formal model of automatic word segmentation, and analyzes the difficulties of word segmentation.

Abstract: Traffic monitoring system based on image processing technology,which can reflect the information of traffic .The cost of use and maintenance is relatively low ,which can be widely applied to monitor complex area and crossing the road. In this paper, a new algorithm for traffic monitoring had be proposed, which combined the SIFT algorithm and kalman filter. Firstly, extracting SIFT targets based on the Pyramid of Gauss difference image, target position was doped out by kalman filter and builded the adapted window. Finally, the SIFT algorithm was used to extract target features and matching them. According to the next kalman location set prediction area, in this area combinated with target center and SIFT features, then selected the best objection for accurate tracking,acquire traffic parameters with traffic fiow and average speed.Experimental results show that detection accuracy is more than 91%,the method has strong robustness and realtime。

Abstract: Local binary pattern (LBP) descriptor could not efficiently describe the gray change in different directions of facial expressions characteristic regions. For this, the directional local binary pattern (DLBP) is put forward to represent facial geometrical characteristic. DLBP encodes the directional information of the face’s facial textures in horizontal, vertical and diagonal three directions, which can effectively describe the characteristic of facial muscles, wrinkles and other local deformation. Experimental results on JAFFE databases demonstrate the algorithm’s effectiveness, where nearly 5 percent recognition rate improvement is obtained beyond traditional LBP. Additional experiments verify robustness and reliability of the proposed DLBP operator within Gaussian white noise and pepper salt noise.

Abstract: This paper proposed a method for detecting fatigue detection Android smart phone system, and applied to the Android system. The system monitors the state of fatigue by smart phones photographs. Face detection method is used to localize the eyes of the driver and the eye region is extracted to monitor the movement of eyelids. An alarm rule is designed based on the PERCLOS standard to detect drowsy driving.Experiments show that the method is accurate rate, running speed, and can be used to monitor driver fatigue during the day.

Abstract: The precautionary assistance system of lane keeping based on monocular vision obtains the road edge features, decides which driving conditions is happened using the fuzzy controller, then gives drivers the corresponding precautionary instructions. In order to improve the processing speed and reduce the needed memory space, the combination of the searching method of connected area and Randomized Hough Transform (RHT) based on two-dimension listing table storage are first proposed in the study, which detects the road edge features. The fuzzy controller is designed with two input and single output, which gives the precautionary instructions in real time. The method proposed in the study is proved that the detection algorithm is simple, and compute velocity is fast, which is fit for the high speed in vehicle driving.