A Comparative Study Of Several Latest Image Retargeting Methods

In this article, firstly, the author made a research to the existing image retargeting technology, and investigated the principle and relative merits of each one. Through the investigation, the author discovered such a fact:With the development of science and technology, many display devices which have a different aspect radio comes to our lives. The traditional image retargeting methods can’t meet the requirements of human beings, so some new technologies of image retargeting are needed to resize the image. But no any kind of image retargeting technology is perfect, that is to say, some methods which have a better result may be failure in some other image. Own laboratory, put forward a kind of image retargeting algorithm based on quadratic programming, and on this basis, the author uses matlab to realize the algorithm, including significant extraction of image and image zoom processing. In addition, to test and verify the effect of this algorithm, the paper also select several other advanced image scaling techniques, they are bilinear interpolation method, the Seam Carving method and the Optimized Scale-and-Stretch method. In order to make this comparison can run smoothly, we use three image retargeting quality standards to guard the comparison. We also discuss the influence of the user’s interactive operation of the three methods. In addition, the author uses the Matlab to implement the Bilinear Interpolation algorithm as a reference to the image retargeting. In these article, the author use Seam Carving method, the Optimized Scale-and-the Stretch method and the Quadratic Programming method to retarget so many images, and choose several ones as the examples, to compare the relative merits of the three methods. Since the high-speed development of image retargeting technology, the image quality assessment method was put out. In order to make the comparison more convincing, this article designed an experiment, using the Kendall harmonious coefficient to prove the consistency of the experimental result and the correctness of the image comparison result. Finally, the author makes a web page using JSP based on the implement of matlab, to retarget the image. That makes it convenient to compare the image retargeting methods.