动画图像目标视觉特征优化检测仿真  

Simulation of Target Visual Feature Optimization of Animated Image

作　　者：孙婷[1] SUN Ting(Department of Dsign and Art, Xi＇an Technological University, Shanxi Xi＇an 710032, Chin)

机构地区：[1]西安工业大学艺术与传媒学院,陕西西安710032

出　　处：《计算机仿真》2018年第5期435-438,443,共5页Computer Simulation

摘　　要：对动画图像的目标视觉特征进行检测,在提高动画图像视觉效果方面具有重要意义。由于动画图像目标的多样性及视觉的流动性,使得动画图像目标视觉特征多样化,且属性不统一。传统特征检测方法主要对动画图像中最感兴趣、最重要的区域,采用同一结构元素对动画图像目标视觉进行膨胀和腐蚀运算,无法有效滤除较大的起伏背景,导致目标视觉特征检测效率低。提出基于改进形态滤波和区域对比的目标视觉特征优化检测方法,在分析动画图像目标视觉特征信息采集系统原理模型的基础上,利用改进的形态学滤波法,根据复杂背景动画图像目标视觉与背景区域的差异性,选择不同的结构元素构造对动画图像目标视觉特征进行预测;采用中央-周边差运算,生成动画图像的特征差异图;利用动画图像的颜色特征和空间特征计算目标视觉特征值,最终得到目标视觉特征。仿真结果证明,所提方法不仅消除了虚警干扰,而且较好地检测出了目标视觉特征。The aim of this article is to overcome defect of traditional detection method of target visual feature of animation image, such as low detection precision. Based on the modified morphological filtering and regional compari- son, an optimization detection method of the target visual feature is proposed. Firstly, based on analyzing principle model of information collection system of the target visual feature, different structural element constructions are selected to predict the target visual feature via the modified morphological filtering according to the difference between complex background target vision and background region. Then, characteristic difference chart of the animation image is generated via central-peripheral difference arithmetic. Feature value of the target vision is calculated via color feature and space feature of the animation image. Thus, the target visual feature is obtained. Simulation resuhs prove that the method eliminates false-alarm interference and detects the target visual feature preferably.

关 键 词：动画 图像 目标 视觉 特征 检测 显著性 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]