Abstract: Completed for the partition of beer bottles convexity characters，an image feature extraction method is put forward based on circular projection.By mass center for the pole，it will transform Cartesian coordinates to a coordinate system.With circular projection calculating，two-dimensional images of characters will transform one-dimensional projection.By using Wavelet Transform（WT） and the Discrete Cosine Transform（DCT），the effective feature extractions of different characters are got.Through the same character similar to the feature extraction experiments，it shows that the method has good anti-noise and very good classification performance.

Key words: beer bottle raised character, circular projection, feature extraction, character recognition

摘要： 针对已分割完成的啤酒瓶凸性字符，提出了一种基于圆周投影变换的图像特征提取方法。以质心为极点将笛卡尔坐标系变换为极坐标系，进行圆周投影计算，将二维的字符图像变换为一维投影。利用小波变换（WT）和离散余弦变换（DCT），提取不同字符的有效特征。通过相同、相似字符特征提取实验，表明该方法具有较好的抗噪声能力和分类性能。

关键词: 啤酒瓶字符, 圆周投影, 特征提取, 字符识别