An Efficient Image Mosaic Algorithm Based on EMD Transform
Yijun Wang, Wei Sen · 2017
This paper proposes an image mosaic algorithm based on empirical mode decomposition (EMD) transform.A complete, fast and efficient EMD image decomposition algorithm is used to decompose the image, and the inverse discrete cosine transform (IDCT) is performed on the result of the transformation.After the interpolation operation, the compressed image is obtained.Then, the Harris algorithm is used to extract the feature points and match these points.The accuracy of the feature point registration is improved by using the triangular area of the first and second adjacent points, and the EMD algorithm is used to compress the image, which greatly reduces the time of image splicing. I.INTRODUCTIONAt present, there are a lot of researches on the key technology of image stitching, image matching and image fusion.However most of the methods in the complex scene image splicing effect is not ideal, and high precision stitching is slowly, time-consuming, poor real-time and can not meet the requirements of fast stitching.The main method of image matching is using C. Harris's Harris operator to detect the feature points [1].Not only the detection speed, but also has the rotation and translation invariance, can achieve sub-pixel level accuracy, and have good robustness.The empirical model is decomposed with a new method for nonlinear and unstable data analysis proposed by NASA's Norden E. Huang in 1998 [2].The two-dimensional empirical mode decomposition has the ability to decompose the image into local narrowband signals.It is born to be used in image processing.At present, there are a variety of attempts base on EMD method in the field of image research [3][4][5][6].In this paper, an image mosaic algorithm based on empirical mode decomposition is presented.The image is subjected to EMD compression processing, and then Harris algorithm is used to stitch the image, and the triangular area of the nearest neighbor feature points is used to remove the pseudo-feature points.To achieve fast, high-precision image stitching. II. IMAGE