Rotational Distortion Analysis And Correction Of Small Scale Image

Image rotation is a geometric transformation in image, and it has been widely used in image processing. There are some errors in the small scale images such as radar target echo images, the character images in the weather facsimile map after rotation. These errors affect the similarity between the rotating image and the original image.This thesis focuses on the analysis of the errors caused by the rotation of the radar target echo image and the character image and the rectification of the rotated image. The errors of the two images are mainly reflected in the shape of the image. The research of the distortion caused by the rotation of the radar target echo image is as follows:The shape can be expressed by the boundary and the boundary curve can be regarded as the connection of many segments. The endpoint of each segment is a turning point on the boundary. Recording the coordinate of turning points by boundary vector, then the boundary of the radar target image after rotation can be obtained by the rotation of the turning points. By observing the deformation point of the image boundary in different rotation angle, the change of their chain code is the same. According to the regulation, the deformation point can be detected and the shape can be rectificated by corresponding rules. The effect of the inpainting is carried out by the smooth degree of the boundary and the similarity of the radar target image before and after the inpainting. The research of the distortion caused by the rotation of the character is as follows:A connected character may be divided into lots of discrete connected domains after the rotation, and broken characters can be judged by marking the connected domain of the character image. Through the observation of broken character image, one pixel or two pixels is found in a pair of breakpoints. Respectively, the possible connection positions between breakpoints are obtained. Two dimensional Fourier description operators are used to extract the feature of the template characters and the rotated characters, and then the distance between feature vectors as the characters similarity is used to obtain the best connection position. Finally, the characters of the inpainting and the characters after the interpolation are compared with horizontal characters respectively, and the effect on recognition is revealed.The experimental results show that the proposed algorithm has good effect on the rectification of the shape distortion caused by small scale image rotation, and is convenient for the subsequent processing of the rotated images.