Research And Application Of Sequence Image Technology

In industrial production and real life, the study based on image sequences has been a hot research field of machine vision,pattern recognition and imageprocessing. The focus of this research are shot through the image sequences, making a separate analysis of each frame image, extracting image feature, and then getting the comprehensive multi frame image of motion object pattern recognition. In finally improving the accuracy of the algorithm and speeding the real-time detection.The modern production technology is changing much rapidly, production efficiency and product quality is the main factor of restricting enterprise development.Especially for the detection of batch production, traditional detection methods and technology has become increasingly unable to afford the requirements of the development of enterprises. In order to improve the detection efficiency and detection accuracy of factory badminton quality,the paper extract the main characteristic as the means, to realize the automation of production quality testing of the badminton through machine vision technology analysis of wind tunnel badminton sequence image,The main research contents of this paper are summarized as follows:1.Upgrade the acquisition platform and the sequence image preprocessing.The upgraded platform can shoot badminton cap, ball head each range flight attitude through an acquisition device. By using prior knowledge and image segmentation, morphological processing,the value of the two moving object extraction,comparison of various restoration algorithm for moving object image, the research select the object appropriate defuzzification algorithm, combined with the map of rotary motion badminton special rotary motion object image reconstruction.2.Build a flight badminton model and analysis the force of the badminton in a wind tunnel.First, badminton coordinate system is established for different angles from the actual flight of badminton,including course coordinates, the coordinate speed sphere, sphere coordinate system, then force analysis is did in these coordinates, the dynamic equation is established, the parameters can be calculated by the equation of each moment and eachposition of the badminton, so as to achieve the purpose of predicting the motion track of the badminton. Finally, studing the current detection badminton wind tunnel testing device, analyzing the badminton performance in different grades of sphere and the cause of the badminton occur different deflection and rotation.we can get the conclusion that the wind tunnel device simulate the feasibility analysis of badminton flight is ok.3. Extracting new characteristic parameters of badminton swing and Improving algorithm. In order to improve the efficiency of badminton online detection, puts forward a kind of more quickly identify16Maojian feature points algorithm. Study the characteristic parameters of badminton head part of the overall stability of the badminton, analysis of the ball head deflection calculation method in horizontal direction and vertical direction and the distance.4, In order to meet the needs of production efficiency, the paper adopts the parallel processing system and uses multi camera image sequence, using multiple CPU core idea of parallel detection of badminton in wind tunnel, given the specific hardware architecture and the software design process of parallel processing programming ideas, shorten the detection time, improve the efficiency of detection system. Finally, design a kind of1VS1SVM on a classifier based on classification for badminton, flight stability and impact level.In this paper,to achieve quality grade badminton tunnel classification detection the algorithm has passed Matlab simulation, and has been programmed in VC6.0. The experimental results show that, the sequence of image processing andanalysis algorithm based on parallel processing system has a certain practical value.