Applied Mechanics and Materials Vols. 220-223

Abstract: Low field nuclear magnetic resonance (NMR) technique was used to evaluate the capacity of NMR spectrometry in characterizing the different behaviors of water protons in navel orange during growth. Carr-Purcell-Meiboom-Gill (CPMG) sequence and mult-exponential regress model were used to calculate the spin-spin relaxation time (T2) of orange samples. Meanwhile, chemical properties (etc. total sugar, acid and solid content) were tested. Statistics analysis shows a correlation between chemical properties of orange and NMR relaxation time during growth, which indicate that interactions between molecules and water protons in orange are involved. Evaluating the behavior of liquid water by NMR shows a new view for better understanding the mechanisms of the macromolecular formation. Changes in NMR relaxation time parameters would be helpful for understanding the mechanism of decay and damage. Therefore, it would be helpful for finding a better food preserving method for the long term storage in the future.

Abstract: C-V model without re-initialization was applied to detect the stone slab contour on simple and complex background respectively. Using the flexible initial curve which is closer to the object boundary, a stable and correct segmentation can be achieved rapidly. For the simple background, this model can exactly detect the stone slab contour. While for the complex background, the result somewhere is an approximation of object contour because the object and background have nearly the same intensity. Additional work will be done to perfect the results.

Abstract: In Order to Improve the Segmentation Effect of the Rice Leaf Disease Images, we Take Optimal Iterative Threshold Method,OTSU Method and Fuzzy C-means Clustering Algorithm to Make Adaptive Segmentation of Rice Disease Images which Were Collected under Different Circumstances. through Comparative Analysis, Experimental Results Show that: Three Methods All Can Effective Separate Spot from the Leaves; in Comparison, the Effect of the Fuzzy C-means Clustering Algorithm Is the Best, but the Number of Iterations Is too many and the Time Spent on it Is the Most; the Effect of OTSU Method Is Lesser, Optimal Iterative Threshold Method Is the Worst. Comprehensive Considering the Segmentation Accuracy and Efficiency, the Paper Chooses OTSU as the Segmentation Method of the Rice Leaf Disease Images.

Abstract: This article introduces a dynamic target simulation device to evaluate dynamic image quality of the aerial camera. The device simulated the positional relationship of the sky and ground in the laboratory when aerial camera flies at high altitude. And it can be used to inspect camera image quality of aerial camera flying at high altitude in laboratory. At a same time, the article focuses on this servo control system of device, building a control model, and giving control strategy. The servo control system was proved stably and reliably after experimental verification. And it provides a strong guarantee for the test of the dynamic image.

Abstract: Model ice-shape measurement is an essential part in the icing test in wind tunnel. The principle investigation of camera calibration and image processing technology based on OpenCV library which apply to ice-shape measurement in wind tunnel is presented in this paper. A kind of software with perfect function and good reproducibility was successfully developed. Ice-shape measurement test was conducted in wind tunnel and the application of OpenCV library in image post-processing was proved to be practical. This program can also be effectively used in aero-optics research, model attitude measurement and model deformation measurement in wind tunnel.

Abstract: In this paper, the idea identify the hook on investment casting shell line based on machine vision has been proposed. According to the characteristic of the hook, we do the image acquisition and preprocessing, we adopt Hough transform to narrow the target range, and find the target area based on the method combining the level projection and vertical projection, use feature matching method SIFT to do the image matching. Finally, we get the space information of the target area of the hook.

Abstract: Lane detection system is generally composed of three modules, lane image acquisition with camera, lane recognition, and lane parameters calculation. The camera setting in lane image acquisition module determines the lane information content in image. This paper proposes the camera extrinsic parameter optimal setting for lane detection, which is derived from the lane projective model in image. Firstly, the general lane projective model in image is introduced. Then the camera extrinsic parameters error influence on the lane parameters calculation is analyzed and the camera extrinsic parameters setting ranges are gotten. Finally, the optimal camera extrinsic parameters settings are derived based on the lane projective model and the compact lane projective model is given. By using the camera optimal setting, the lane information content in the images captured by the camera can be maximized and the lane projective model can be compacted.

Abstract: Blind deconvolution is the restoration of original image from a blurred one when the blur kernel is unknown. While recent algorithms have afforded dramatic progress, the results are still far from perfect in terms of efficiency and stability. In order to gain a stable, unique and effective solution, this paper uses a scale invariant and sparse regularization function to exert regularization constraints on the original image and PSF simultaneously. An experiment is conducted to verify that our image blind recovery algorithm is robust and has stable convergence.

Abstract: This paper studies a high-precision measuring method about minimum thickness of thin neck and bending rigidity of flexible bar based on machine vision. The measurement system combines with the advantages of image vision inspection and laser displacement detection. After rough leveling and precise leveling to the clamped workpiece through the interworking of the camera, laser displacement sensor and servo motor, the image of thin neck and the upper profile of workpiece are captured by camera and laser displacement sensor respectively. Then the workpiece is rotated 180 through servo motor so as to capture the image of thin neck again and detect the bottom profile. Based on the two groups of images and profiles acquired above, we can figure out the minimum thickness of thin neck and bending rigidity of the workpiece with high precision. The effectiveness and feasibility of this method are confirmed by computer simulation with ANSYS software and experimental results with needle gauge and flexible bar.

Abstract: Vibratory feeding is the most significant approach for parts orientation and feeding in production; the process of production automation has been seriously hindered due to the problems with low feeding efficiency, high jamming rate. This project combines machine vision and vibratory feeding technology in substitute for the currently used mechanical orientation mechanism, in order to automatically recognize work pieces of different size, shape and color. The special vibratory feeder and frame are designed. According to the system work requirement, in the thesis, the adjustable light source and illumination system are designed, and the CCD camera and optical lens are chosen. The pneumatic actuator is also designed with the microcomputer as the controller. The area, perimeter, dispersion and invariant moment parameters are used in parts type identification. The image centroid is used to confirm parts position. By using the Hough transform, the image edge can be extracted to get the deflection angle. By using Matlab, the image can be captured, processed, analyzed and judged, meanwhile, communicate with actuator and control it.