An Application of Least Squares Method for Image Process

• Main Authors: Shih-WeiYu, 于世偉
• Other Authors: Shih-Yu Shen
• Format: Others
• Language: zh-TW
• Published: 2019
• Online Access: http://ndltd.ncl.edu.tw/handle/buug6a

碩士 === 國立成功大學 === 數學系應用數學碩博士班 === 107 === This thesis is to explore how to obtain the affine transformation relationship from two similar images by using the least squares method and the directional derivative to calculate the image, and then compare the images. The comparison method is one of them. The image is subtracted from the other image by the obtained affine transformation relationship. The subtraction is performed by taking the RGB value of each pixel of the image from the first point to the last point of the image. Subtract in order. In the second chapter, when there are more than three corresponding conjugate pairs between two images, the conjugate pairs are used to find the conversion relationship between the two images, and finally the least squares method of affine transformation is proved. However, if these conjugate pairs are still able to obtain the conversion relationship, the next experimental comparison does not require a conjugate pair, but is compared in a certain area of the two images. In the third chapter, the experimental process is to convert a standard image into an affine and convert it into another image. In this case, the affine transformation is set to be discussed in the case of translation and rotation, that is, in the known conversion relationship. In the actual comparison of the fourth chapter, two images are actually taken, and the discussion is made in the case where the conversion relationship is unknown, and the results are described in detail in this paper.