计算机工程与设计
計算機工程與設計
계산궤공정여설계
COMPUTER ENGINEERING AND DESIGN
2010年
7期
1519-1522
,共4页
阴影去除%互相关函数%交叉熵函数%运动检测%运动跟踪
陰影去除%互相關函數%交扠熵函數%運動檢測%運動跟蹤
음영거제%호상관함수%교차적함수%운동검측%운동근종
shadow elimination%CCF%CEF%moving detection%moving tracking
为了抑制运动人体检测与跟踪中运动人体阴影的影响,提出了一种基于光照模型和图像纹理特征的阴影检测算法.首先依据光照模型理论,将互相关函数引入算法,并根据背景图像和当前图像对应像素点的互相关函数值判断其是否为阴影点.在此基础上,依据图像的纹理特征,将交叉熵函数引入算法,对互相关函数判断为非阴影点的像素,进行二次判断.该算法利用泰勒级数将交叉熵函数中的对数和除法运算转化为乘法运算,简化了运算过程.通过白天和晚上两种条件下的实验,验证了算法的准确性和有效性.
為瞭抑製運動人體檢測與跟蹤中運動人體陰影的影響,提齣瞭一種基于光照模型和圖像紋理特徵的陰影檢測算法.首先依據光照模型理論,將互相關函數引入算法,併根據揹景圖像和噹前圖像對應像素點的互相關函數值判斷其是否為陰影點.在此基礎上,依據圖像的紋理特徵,將交扠熵函數引入算法,對互相關函數判斷為非陰影點的像素,進行二次判斷.該算法利用泰勒級數將交扠熵函數中的對數和除法運算轉化為乘法運算,簡化瞭運算過程.通過白天和晚上兩種條件下的實驗,驗證瞭算法的準確性和有效性.
위료억제운동인체검측여근종중운동인체음영적영향,제출료일충기우광조모형화도상문리특정적음영검측산법.수선의거광조모형이론,장호상관함수인입산법,병근거배경도상화당전도상대응상소점적호상관함수치판단기시부위음영점.재차기출상,의거도상적문리특정,장교차적함수인입산법,대호상관함수판단위비음영점적상소,진행이차판단.해산법이용태륵급수장교차적함수중적대수화제법운산전화위승법운산,간화료운산과정.통과백천화만상량충조건하적실험,험증료산법적준학성화유효성.
To suppress the impact of the shadow of the moving body in detecting and tracking moving targets, a shadow detection algorithm based on illumination model and image texture characteristic is presented. Firstly, the cross-correlation function (CCF) based on illumination model is involved, and whether a pixel is a shadow point is determined by its value of the CCF. Then, the cross-entropy function (CEF) based on image texture characteristic is proposed. CEF is used to evaluate a non-shadow pixel judged by CCF. According to the Taylor series, logarithm and division in CEF are transformed into multiplication. Finally, by making experiments under day and night conditions, the validity and efficiency of the presented algorithm is demonstrated.