摘要
体育运动员在高强度运动时,身体各部位会出现一定的损伤,在进行治疗过程中,需通过采集损伤图像对图像中损伤部位进行识别分析,当前相关方法存在识别精度低、效率差的问题,提出基于鱼群算法的高强度运动损伤图像破损部位识别方法。对运动损伤图像实行灰度转换操作,根据自适应阈值化与数学形态结合提取图像中损伤部位轮廓。选取种子点,利用活动轮廓向初始轮廓逼近,采用曲线拟合法对获取的离散点拟合,得到最终破损轮廓。利用提取到的破损位置像素点数量及相关信息构建数字矩阵,将图像按照列串接的模式排为长度是64的特征向量,计算图像总体均值向量,将计算结果、训练样本以及待识别图像样本代入欧氏距离中,得到运动损伤的图像破损部位初步识别结果。利用鱼群算法以聚类的方式实现图像分割,采用聚类分割结果彩色描绘原始图像像素类别,并计算运动损伤图像中相对破损比例面积,识别高强度运动损伤图像破损部位。实验结果表明,采用上述方法可提高运动损伤图像破损部位识别的准确度,缩短了识别时间,在确定运动损伤部位方面具有一定的可行性。
It is necessary to identify and analyze the injury part in image by collecting the injury image. Currently , methods lead to low recognition accuracy and low efficiency. Therefore, a method to identify the injury part in high-intensity sport injury image based on fish swarm algorithm was put forward. Firstly, the gray-level transformation was performed on the sport injury image. Secondly, adaptive thresholding was combined with mathematical morphology to extract the contour of injury part. Thirdly, seed points were selected and the active contour was used to approximate the initial contour. Fourthly, the curve fitting method was used to fit the obtained discrete points. Then, final broken contour was obtained. Fifthly, number of pixel points of injury part and related information were used to construct the digital matrix. According to the pattern of tandem connection, images were arranged into a feature vector whose length was 64. Then, mean vector of image was calculated. The calculated results, training samples and image samples to be recognized were introduced into Euclidean distance, so as to get initial recognition results of injury part in image. Moreover, fish swarm algorithm and clustering method were used to realize image segmentation. Finally, the clustering segmentation result was used to describe original image pixel category. Meanwhile, the relative broken proportion in the sport injury damage image was calculated. Thus, damaged parts of high-intensity sport injury image were recognized. Simulation results show that the proposed method can improve the accuracy of recognizing damage part of sport injury image, shorten the recognition time. Meanwhile, this method has certain feasibility in determining the injury part.
作者
王光明
WANG Guang-ming(School of Physical Education, Pingdingshan University, Pingdingshan Henan 467000, China)
出处
《计算机仿真》
北大核心
2019年第8期412-416,共5页
Computer Simulation
基金
河南省农村城镇化进程中体育公共服务供给改革研究(2016CTY019)
关键词
高强度
运动损伤图像
破损
识别
High strength
Sport injury image
Damaged
Recognition