关键词: 形状描述, 特征提取, 形状识别

Abstract:

Shape feature is an important visual feature of image, and its extraction method is a hot topic in the field of shape recognition, image retrieval and image matching. Spike parameter is used to reflect the angle change of wear particle contour. The larger of the spike parameter, the wear particle is more shaper, and its abrasive action is bigger. Based on the spike parameter, four spike functions that used for shape feature extraction are proposed. The spike-angle function and spike-height function are used to extract the detail features of shape contour, the spike-area function and spike-distance function are used to extract the global features of shape contour. When extracting shape features by spike functions, multi-step spike-angle function and spike-height function are used, and single-step spike-area function and spike-distance function are used. In order to eliminate the effect of the starting point to the calculating of spike function, the amplitudes of the normalized Fourier transform coefficient of multi-scale spike function are used as shape features. The proposed method is verified in the MPEG-7 and Swedish leaf datasets. Compared with other methods, the results show that the proposed method can reach the higher correct classification rate, and has stronger anti-noise ability.

Key words: shape description, feature extraction, shape recognition