Hand gestures are powerful means of communication among humans and sign language is the most natural and expressive way of communication for dump and deaf people. In this work, real-time hand gesture system is propose...Hand gestures are powerful means of communication among humans and sign language is the most natural and expressive way of communication for dump and deaf people. In this work, real-time hand gesture system is proposed. Experimental setup of the system uses fixed position low-cost web camera with 10 mega pixel resolution mounted on the top of monitor of computer which captures snapshot using Red Green Blue [RGB] color space from fixed distance. This work is divided into four stages such as image preprocessing, region extraction, feature extraction, feature matching. First stage converts captured RGB image into binary image using gray threshold method with noise removed using median filter [medfilt2] and Guassian filter, followed by morphological operations. Second stage extracts hand region using blob and crop is applied for getting region of interest and then “Sobel” edge detection is applied on extracted region. Third stage produces feature vector as centroid and area of edge, which will be compared with feature vectors of a training dataset of gestures using Euclidian distance in the fourth stage. Least Euclidian distance gives recognition of perfect matching gesture for display of ASL alphabet, meaningful words using file handling. This paper includes experiments for 26 static hand gestures related to A-Z alphabets. Training dataset consists of 100 samples of each ASL symbol in different lightning conditions, different sizes and shapes of hand. This gesture recognition system can reliably recognize single-hand gestures in real time and can achieve a 90.19% recognition rate in complex background with a “minimum-possible constraints” approach.展开更多
树叶晃动、光照变化等自然场景下的动态背景会影响运动目标检测的准确性,区分动态背景和前景目标的变化是复杂场景下运动目标检测的首要任务。针对现有的前景提取算法逐点提取前景从而导致计算资源浪费的问题,提出了一种区域提取与改进L...树叶晃动、光照变化等自然场景下的动态背景会影响运动目标检测的准确性,区分动态背景和前景目标的变化是复杂场景下运动目标检测的首要任务。针对现有的前景提取算法逐点提取前景从而导致计算资源浪费的问题,提出了一种区域提取与改进LBP(Local Binary Patterns)纹理特征相结合的运动目标检测算法。首先,将图像分为大小相等的图像块,利用各图像块的统计特性建立核密度估计(Kernel Density Estimation,KDE)模型,并用KDE模型估计出前景区域。然后,计算前景块中所有像素点的改进LBP纹理特征直方图。最后,通过直方图匹配提取所有的前景像素实现目标的精确提取,并用概率模型更新背景。实验结果表明,该方法在快速提取运动目标前景区域的同时能够消除大部分动态背景产生的干扰,相比传统算法更适用于自然场景下的运动目标检测。展开更多
文摘Hand gestures are powerful means of communication among humans and sign language is the most natural and expressive way of communication for dump and deaf people. In this work, real-time hand gesture system is proposed. Experimental setup of the system uses fixed position low-cost web camera with 10 mega pixel resolution mounted on the top of monitor of computer which captures snapshot using Red Green Blue [RGB] color space from fixed distance. This work is divided into four stages such as image preprocessing, region extraction, feature extraction, feature matching. First stage converts captured RGB image into binary image using gray threshold method with noise removed using median filter [medfilt2] and Guassian filter, followed by morphological operations. Second stage extracts hand region using blob and crop is applied for getting region of interest and then “Sobel” edge detection is applied on extracted region. Third stage produces feature vector as centroid and area of edge, which will be compared with feature vectors of a training dataset of gestures using Euclidian distance in the fourth stage. Least Euclidian distance gives recognition of perfect matching gesture for display of ASL alphabet, meaningful words using file handling. This paper includes experiments for 26 static hand gestures related to A-Z alphabets. Training dataset consists of 100 samples of each ASL symbol in different lightning conditions, different sizes and shapes of hand. This gesture recognition system can reliably recognize single-hand gestures in real time and can achieve a 90.19% recognition rate in complex background with a “minimum-possible constraints” approach.
文摘树叶晃动、光照变化等自然场景下的动态背景会影响运动目标检测的准确性,区分动态背景和前景目标的变化是复杂场景下运动目标检测的首要任务。针对现有的前景提取算法逐点提取前景从而导致计算资源浪费的问题,提出了一种区域提取与改进LBP(Local Binary Patterns)纹理特征相结合的运动目标检测算法。首先,将图像分为大小相等的图像块,利用各图像块的统计特性建立核密度估计(Kernel Density Estimation,KDE)模型,并用KDE模型估计出前景区域。然后,计算前景块中所有像素点的改进LBP纹理特征直方图。最后,通过直方图匹配提取所有的前景像素实现目标的精确提取,并用概率模型更新背景。实验结果表明,该方法在快速提取运动目标前景区域的同时能够消除大部分动态背景产生的干扰,相比传统算法更适用于自然场景下的运动目标检测。
