儿童摔倒行为图像检测数学建模方法仿真.pdfVIP

第33 卷第11 期 计算机仿真 2016 年 11 月 文章编号:HX16-9348( 2016) 1l-0452-04 儿童摔倒行为图像检测数学建模方法仿真 王艳 (长春财经学院，吉林长春 13仪削) 摘要:由于儿童对周围环境认知不健全，使得行为存在很多不确定性。传统的儿童摔倒行为图像检测过程中，主要通过采集 儿童摔倒的前提行为特征进行建模分析，忽略了所处环境造成的不确定性对建模分析的影响，导致建模分析不准确的问题。 为解决上述问题，提出基于仿射变换的儿童摔倒行为图像检测数学建模方法。通过背景减法获得儿童摔倒行为差值图像， 结合VaIly-Emphasis 自适应阙值理论对差分图像进行二值化处理，并以灰度级作为分割图像的阀值，获得儿童摔倒行为差 值图像通过自适应二值化后的前景图像，并组建六参数的仿射变换模型，利用上述模型估计出多帧儿童摔倒行为图像在下 一时刻的匹配背景图像，将背景配准后的图像进行双向差分相乘，计算出儿童目标的质心位置，完成儿童摔倒行为图像检 测。实验结果表明，所提方法能够有效提高儿童摔倒行为图像检测准确率，具有较好的鲁棒性和实时性。 关键词:儿童摔倒行为:图像检测:数学建模 中图分类号:1凹91. 41 文献标识码:B A Child Acts Fall Image Detection Method of Mathematical Modeling Simulation WANG Yan (Changchun University of Finance and Economics , Changchun Jilin 130000 , China) ABSTRACT :The environment cognition of children is unsound. It makes the behavior have many uncertainties. In this paper , we propo能d a mathematical modeling method for image detection of children tumhle based on 由e affine U田四formation. Firstly ，由e author obtained the difference image of children tumhle through background subtraction and conducted a binarization pn陀essing to difference image integrated with Vally-Emphasis self-adaption threshold value. 币le gray level was also used as the threshold value of segmented image to obtain the foreground image of chil- dren tumhle difference image after self-adaption binarization processing. Then 也e affine transfo口nation model with six parameters was built and the above model was used to estimate the matching background image of mul