The Intelligent Storagealgorithm Based On K-means

Although video monitor technology brings us convenience, massive videos producedby video monitor make it a disaster to store and view videos, so it is significant to present aintelligent store algorithm. intelligent store technology can cut the sequence of the videointo several parts, and makes it convenient for customers to view the video and process thevideo。Because intelligent store technology will ignore pure backgrounds and constantpictures, it not only saves the storage fees, but also makes the sequence of the video simple.Based on Anomaly detection of camera, clustering algorithm is used to classify a set offrames, stores them into labels prepared and forms the standard video file during themonitoring cycle. This process involves the extraction of dynamic feature, the use ofclustering algorithm for clustering and classified storage. Now I will present the main workof the paper:(1) Dynamic multi-feature modelIn order to describe the character of video frame, this paper uses anomaly detectionalgorithm based on the difference of histogram to solve the dynamic multi-feature model.Numerical results demonstrate the character based on the difference of histogram can makea distinction between abnormal video sequence and normal video sequence.(2) Utilizing the K-means clustering algorithm for the first time, the frames have beenclassified This paper introduces K-means clustering algorithm into intelligent storage forthe first time, it separates video frames into several parts. Then, according to the sequenceof the video it stores some necessary video frames into labels prepared, while quits othersin order to accomplish the purpose of intelligent storage.(3) The method of the weighted distance based on the information gain ratioIn the K-means clustering algorithm, the distance is a method to measure the similarity.The method of the weighted distance based on the information gain ratio can reflect thesignificance of the character component in the classification, so it can promote the qualityof clustering.Intelligent Storage algorithm based on K-means is different from traditional ways tostore, it not only solves the problem of be and not be. But it also accomplishes the goal ofclassification.