The Analysis Of Paddle Frequency On Kayaking (200m) Hydrostatic Sport

Kayaking (200m) hydrostatic movement was accepted as an event officially in2012London Olympics，Previously, as a non-Olympic events，Kayaking(200m) hydrostaticproject did not receive much attention, so we short of the project technical information.in order to get high grades, we need to measure, record and even realize the paddlefrequency data’s comparison.In this article, we analyze the kayaking (200m) hydrostatic movement From theperspective of machine vision, Propose a set of analysis programs。The main algorithmsare: kayaking positioning, kayaking tracking, paddle positioning and key frame grabber,the main contents are as follows:(1) kayaking positioning：Based on the detection of the canoe kayak detection on theprevious chapters, The canoe was tracked by video sequence continuity.Due to the needto analyze the details of the video sequence, while position the kayak, as moving targetsto be tracked own characteristics, environment complexity of the position algorithm, howto get a better position effect has always been a hot topic. This paper study the Kayakdetection under the static background，we Introduces several classical target detectionmethods, On the basis of analyzing the classic target detection method and detailing thepopular algorithm of PCA to locate kayaking detection, According to the dynamic noiseof image is mainly concentrated in the surface, propose combining the frame differencemethod and kayaking horizontal direction and the vertical direction integral projection tolocation the image boundary position of kayaking，through the contrast experiment,wevalidate its desirability.(2) kayaking tracking: Briefly introduced target tracking algorithm, Mean Shiftalgorithm is a tracking algorithm that can find local optima by estimating the probabilitydensity of the data. Due to it is simple to use and fast to match, we select Mean Shift askayaking tracking method, Mean Shift algorithm uses grayscale or color characteristicsof the target as the feature to build models, and therefore, when there exists some color inthe background that is similar to the color or grayscale of the target, the detecting performance of the algorithm will be affected. In those cases with complex backgrounds,there is always a gray-scale or color distribution that is similar to the tracking targets. ButCross correlation is not affected by the difference between contrast and brightness, Tosolve such problems, this thesis analyzes the basic principle of Mean Shift trackingalgorithm, combining with the cross-correlation information，and due to the complexityof computational cross-correlation information, using fast Fourier transform processing,can help to improve target tracking feature matching degree, and reduce the appearanceof false matches in complex backgrounds.(3) paddle positioning and key frame grabber: On the basis of tracking the kayak,the last step is to complete the kayaking paddle-frequency analysis, combining geometricfeatures and contour information positioning paddle, According to a recent distancebetween the paddles and kayak hull, The canoe is divided into the paddle contact withwater instantly or not two categories. By calculating the athletes the paddle contact withwater instantly time,Realize the kayak paddle frequency analysis.