形状特征描述在基于内容的图像检索与识别研究中具有重要地位,文中对图像检索及识别中常用的形状描述进行了介绍。提出一个新的基于区域中心分布的方案来对图像形状特征进行描述,该描述子以待识别对象的区域二值图像的中心点为圆心,将各点到圆心的距离进行规范化处理,统计落入以区域中心为圆心的各环内的点数与总像素点数n的比例,从而形成一个基于中心分布的形状特征向量。经过数学证明该描述方案提取的形状特征具有缩放、旋转和平移不变性。文中通过使用该方案提取的图像特征进行图像识别检索实验,取得了理想的实验结果,说明了它是一个有效的图像区域形状描述子。
The shape characteristic description has an important position in image identify research according to the picture's contents, In this paper, introduce the shape descriptor in figure retrieve and recognition. A scheme based on center distributing of the region was present to descript figure shape feature. Unitize distances among pixels to the center point of the region and then calculate proportions of the pixel number in different rings to the total number. The different ring has a different proportion. With these proportions,get a shape figure vector. This scheme presented can bring the figure shape feature with zoom invariability character, rotation invariability character and scale invariability character. Finally, had an experiment and proved that the ,scheme presented is effective.