论文部分内容阅读
基于内容的机械设计图像检索对机械设计行业有重要作用。提出一种基于几何特征组合的机械设计图像检索方法。通过机械设计图像形状区域对应的面积、周长等信息,得到7个具有不变性的几何特征如密实度、矩形度等,并将7个特征进行组合,构成机械设计图像的形状特征描述符,用于机械设计图像检索。该形状特征描述符具有易于计算和无需归一化的特点。实验表明,该方法在机械设计图像检索中的性能优于傅里叶描述符方法和Hu不变矩方法。
Content-Based Mechanical Design Image retrieval plays an important role in the mechanical design industry. A mechanical design image retrieval method based on geometric feature combination is proposed. Seven geometric features with invariance, such as density and squareness, are obtained through mechanical design of the image area corresponding to the area and circumference, and the seven features are combined to form the shape descriptor of the mechanical design image. Used for mechanical design image retrieval. The shape descriptor has the characteristics of easy calculation and no need of normalization. Experiments show that this method is superior to Fourier descriptor method and Hu invariant moment method in mechanical design image retrieval.