The developing of multimedia to make the video data to increase very quickly.So how to acquire the data that we want in a short time is a more important topic. Shot change detection is the first step for latter operation like classification and annotations. There are two type of shot change, one is abrupt shot change and the other one is gradual transition. Dissolve is the one of gradual transition that often seen but hard to detection, so in the paper would to propose a robust method to solve this problem. In this paper we use fractal orthonormal basis for our feature to compare frames in the video to the first frame of video, and use the quantification between those frames to draw a graph. By analyzing the graph and the characteristic of dissolve in the graph we can locate the approximately the start frame and the end frame of the dissolve. But by the action of video camera or motion of object in frame we may obtained the inaccurate start frame or end frame of the dissolve. So we need to refine the more accurate start and end frame of the dissolve, and we will explain about this in Chapter 3-4

Digital multimedia can be distributed via the internet efficiently with superior compression technologies. The chance of distributing digital intellectual properties, such as image, music, films, and software, being large-scale unauthorized copied and distributed are much increasing one possible and practical solution for the copyright protection is information hiding technology. Information hiding technology embeds a special data into multimedia data for copyright protection. However, the embedded data may be damaged by malicious attacks or common signal processing. In this thesis, an information hiding technique based on Fractal Orthonormal Basis is proposed. First, the original image is divided into NxN Range blocks, each range block is substituted by several Domain blocks (Fractal Orthonormal Basis), then the watermark information is embedded into the coefficients of the fractal orthonormal basis. Besides, our technique will be compare with the other two watermarking algorithm (using DCT and DWT). After the attacks of cropping, down-scaling, median filter, smoothing, noise, JPEG, SPIHT and EZW compression, the Fractal Orthonormal Basis watermarking technique shows better result of capacity, transparency and robustness. In addition, we only store parts of compression fractal codes and the permutation seed, and these can be the secret key for the security.

The objective of the present work is to propose a novel method to extract a stable feature set representative of image content. Each image is represented by a linear combination of fractal orthonormal basis vectors. The mapping coefficients of an image projected onto each orthonormal basis constitute the feature vector. The set of orthonormal basis vectors are generated by utilizing fractal iterative function through target and domain blocks mapping. The distance measure remains consistent, i.e., isometric embedded, between any image pairs before and after the projection onto orthonormal axes. Not only similar images generate points close to each other in the feature space, but also dissimilar ones produce feature points far apart. The above statements are logically equivalent to that distant feature points are guaranteed to map to images with dissimilar contents, while close feature points correspond to similar images. In this paper, we adapt the Multiple Instance Learning paradigm using the Diverse Density algorithm as a way of modeling the ambiguity in images in order to learning concepts used to classify images. A user labels an image as positive if the image contains the concepts, as negative if the image far from the concepts. Each example image is a bag of blocks where only the bag is labeled. The User selects positive and negative image examples to train the concepts in feature space. From a small collection of positive and negative examples, the system learns the concepts using them to retrieve images that contain the concepts from database. Each concept having similar blocks becomes the group in each image. According groups¡¦ location distribution, variation and spatial relations computes positive examples and database images similarity.