Advanced Materials Research Vols. 694-697

Abstract: In surveillance system,it was challenging to improve real-time in the presence of dynamic background motions.We presented a real-time algorithm for foreground-background segmentation based on codebook model.Pixels were converted from RGB space to YCrCb space,background model used layered model.Firstly we established a basic codebook background model and then got rough background pixels by twice frame difference,and then only trained rough background pixels which have removed foreground pixels.Secondly the foreground was segmented from the background and we updated the background in real-time. The experimental results show that this method can save time of establishment of codebook background model and has small calculation and high accuracy in scenes such as illumination changes,swaying trees and stopped objects should be considered part of the background objects.

Abstract: With the rapid development of computer science, image processing, and pattern recognition, people are paying more and more attention on research and application of machine vision, which has resulted in excellent results in many fields. Depending on the features of untouched, fast-speed, real-time, proper precision, anti-interference, the inspection technology based on machine vision has been researched deeply and comprehensive applied aboard. Recently, the size measurement technology based on machine vision has been applied in the size measurement of the machine processing parts. A case study of typical regular mechanical parts, algorithm of position and size measurement of line, circle center and radius are researched, and good results are got, which established an academic foundation for further measurement of complex parts.

Abstract: One framework that fuses features for abnormal behavior recognition in videos was proposed, which provides increased robustness to noise and pose variation. First, ROI division of suspected pedestrian is achieved by optical flow method, which greatly reduces the dimensions of training data. After that, the proposed method uses the adaboost training to recognize pedestrian effectively. To capture the imaging variations and attributes of individuals, we use types of features: center track and inclination angle. Finally, D-S evidence theory is used to combine these features, which aims to recognize the abnormal behavior correctly. The results demonstrate that the recognition rate can be improved by the fusion of features.

Abstract: The system hardware and software architecture of smart camera does not follow a common standard subject to their versatile application field. This paper presents architecture of smart camera in order to making the smart camera unified and applied to typical fields. It provides a hardware and software case to inspect the steel ball quality in the real time. The result shows that the architecture can meet the requirements of smart camera in several application fields.

Abstract: The robots identify, locate and install the workpiece in FMS system by identifying the characteristic information of target workpiece. The paper studied the recognition technology of complex shape workpiece with combination of BP neural network and Zernike moment. The strong recognition ability of Zernike moment can extract the characteristic. The good fault tolerance, classification, parallel processing and self-learning ability of BP neural network can greatly improve the accurate rate of recognition. Experimental results show the effectiveness of the proposed method.

Abstract: The hand made beaded pad is getting more and more expensive due to the increasing labor costs and its necessary to develop the beaded pads weaving machine. One of the key problems for beaded pads weaving machine developing is the beaded pad image pattern recognitions including the beads number and colors. This paper focuses on the researches of identifying the beads number and the colors of the beaded pad image. First, the beads number of the pad image is identified by threshold and calibration methods. Then, the beads are separated from the image background and the beads colors are identified by the characteristics analysis for the RGB color space of the pad image. Finally, the verification is further provided by computer simulation and the beads pad reconstruction, and the results show that this proposed method can identify the number and the colors for the beads in pad image accurately.

Abstract: Automatic road extraction from remote sensing imagery is one of the hot topics in the field of remote sensing. surveying and mapping. computer vision. etc. In this paper. we summarize the research status of road extraction from high resolution remote sensing imagery. The difficulties and trends of the research are analyzed in the paper.

Abstract: The paper proposes a new method for moving objects detection based on fusion of three frames differencing and Gaussian Mixture Model (GMM). In the method, two images are obtained by three frames differencing, then the adaptive background are modeled and updated by GMM for each pixel in the two differencing images. Next, two differencing images are done logic "and" operation to get the shape of the moving object. Finally adopt the mathematical morphology operation to eliminate noise and the small areas of non-objects motion parts. The simulation results show that the proposed method can detect the objects effectively and real-time. So it can be applied in visual surveillance system effectively.

Abstract: Researched on weft fiber cut problems of glass fiber, improved the efficiency of textile production. Glass fiber textile machine is a major producer machine of glass fiber cloth. Textile machines weft detection usually uses the contact type in production, requires that the weft maintains certain pressure to the sensor. Using this method will cause glass fiber weft bristling, and will produce glass fiber floating dust. Damage to the textile machine and has the harm to the human body health. Used video surveillance method to detection the weft, image recognition and speed directly affects the stability of the system. This paper presented a detection methods of glass fiber textiles weft fiber cut based on neural network-based, selected multiple features which were directly related to the image with the weft as neural network input vector, through repeated training samples to remove tiny ripple effects which were caused by weft textile jitter, overcome the traditional method detection accuracy was not high. Experimental results show that this method can effectively avoid the weft jitter, making accurate detection of the weft fiber cut, and achieved satisfactory results.

Abstract: Aimed at Gaussian white noise, a video noise estimation algorithm based on block neighborhood relevance is demonstrated. Firstly, a differential operator is taken between two sequential video frames. Then, the smooth blocks are selected between original video and differential video based on block neighborhood relevance. Finally, by computing the weighted average of the noise variance of the smooth blocks, the noise variance estimation is achieved. Experimental results show that the proposed algorithm works well.