根据成像原理，分析出图像失真主要来源于径向畸变和透视失真；建立面阵CCD （电荷耦合元件）相机采集轴侧表面图像的专用成像模型，提出了快速展开拼接算法，获得检测图像和展开复原图像的坐标对应关系，采用双线性插值法对图像进行展开复原；利用相位相关法获得相邻图像的平移位置关系，将多张展开图像进行轴向和径向拼接，生成完整的侧面展开图。试验结果表明：将6幅轴的侧面缺陷图像展开拼接成分辨率为1 624像素×750像素的矩形图像所需时间为1.793 s，每幅图像展开的平均时间为0.090 s，展开拼接图像中缺陷几何特征参数的相对误差在3.9%以内。这满足了轴类工件表面无损检测对实时性和准确性的要求。

According to the imaging principle, the analysis shows the imaging distortion is from the radial and perspective. A special imaging model about the surface image of the CCD area-array camera is established. A fast spreading and splicing algorithm is proposed. The relationship between the original image and the reconstructed image is obtained. The image is reconstructed by bilinear interpolation method. Phase correlation method was used to get translational information and splice images in two ways, and the complete axle workpiece spreading images were created. The experimental results indicate that it takes 1.793 s to spread and splice 6 axle surface images into a rectangular image with the resolution of 1 624 pixel×750 pixel. The average time of each spreading is 0.090 s. The relative errors of the defects' geometric characteristic are less than 3.9%. The method could satisfy the real time and accuracy requirements of the axle workpiece's surface detection.