低比特率下篮球视频图像自适应监测技术

Basketball Video Image Adaptive Monitoring Technology with Low Bit Rate

当前利用人工的方式对篮球视频图像进行监测时,存在图像采集效果不理想、所得图像噪声大及图像分类准确性低等问题,无法得到高效的监测结果。采用线阵摄像机、图像采集卡、摄像头、照明灯光结合的结构实现篮球视频图像的采集,其中的镜头型号为尼康Nikkor 28 m mf/2. 8D,固定焦距是28 mm,支持光圈自动调整,可以高效采集篮球视频图像;对所得图像进行线性灰度变换操作;并分别采用均值、中值与同态滤波,实现篮球视频图像的去噪;利用篮球图像运动轨迹骨架裁剪的方式,正确将图像进行连接,接着对篮球图像参数进行计算;利用三角网的构建方式——凸壳插值法,实现图像的特征提取,通过得到的三角网,获得图像结构散点,实现三角剖分,完成监测。实验结果表明,该方法不仅可以增强图像采集效果和分类精度,还能很好地控制图像中的噪声。

At present,when using artificial method to monitor basketball video image,there are some problems such as poor image acquisition effect,large noise and low accuracy of image classification,which can not get high efficiency monitoring results. Basketball video image acquisition is realized by the combination of linear camera,image acquisition card,camera and lighting. The lens model is Nikon Nikkor 28 m mf/2.8 D,with a fixed focal length of 28 mm,which supports automatic adjustment of aperture. The basketball video image can be collected efficiently,and the linear gray level transform operation is carried out,and the mean value,median value and homomorphism filter are used to realize the result. The basketball video image denoising;using the basketball image track skeleton cutting way,the image is connected correctly,then the basketball image parameters are calculated;using the triangular network construction method-convex shell interpolation method,The feature extraction of the image is realized,and the triangulation is realized through the obtained triangulation network and the scattered points of the image structure. The experimental results show that the proposed method can not only enhance the image acquisition effect and classification accuracy,but also control the noise in the image.