Jae-Gark Choi

Adaptive video-dissolve detection method based on correlation between successive scenes

We establish the importance of correlation between successive scenes in dissolve detection and propose a new adaptive dissolve detection method based on the analysis of a dissolve modeling error, which is the difference between an ideally modeled dissolve curve with no correlation and an actual dissolve curve including correlation. The proposed method consists of two steps. First, candidate dissolve regions are extracted using the characteristics of a downward-convex parabola; then each candidate region is checked using the dissolve modeling error. If the dissolve modeling error for a candidate region is less than a threshold equal to the dissolve modeling error with a target correlation, the candidate region is identified as a dissolve region with a lower correlation than the target correlation. The threshold is determined adaptively from the variance at the start and end of the candidate region and the given target correlation. By considering the correlation between successive scenes, the proposed method is able to function as a semantic scene-change detector. The proposed algorithm was tested on various types of data, and its performance proved to be more accurate and reliable than that of other, commonly used methods.

RST-Resistant Image Watermarking Using Invariant Centroid and Reordered Fourier-Mellin Transform

This paper proposes a new image watermarking scheme which is robust to RST attacks by improving Fourier-Mellin transform based Watermarking (FMW). The proposed scheme reorders and modifies function blocks of FMW for improvement of realization and performance. Unlike FMW, our method uses Log-Polar Map (LPM) in the spatial domain for scaling invariance, while translation invariance is provided by the use of an invariant centroid as the origin of LPM. Invariant centroid is a gravity center of a central area on gray scale image that is invariant although an image is attacked by RST. For this, its calculation method is proposed. Also since LPM includes the property which transforms rotation of Cartesian coordinates system into a cyclic shift, 2-D DFT is performed on the LPM image and the magnitude spectrum extracted to provide a domain that is rotation invariant. The resulting domain, which is invariant to RST, is then used as the watermark-embedding domain. Experimental results demonstrate that the proposed scheme is robust to RST attacks.

Correlation based video-dissolve detection

We propose a dissolve detection method based on the analysis of a dissolve modeling error that is the difference between an ideally modeled dissolve curve without any correlation and an actual variance curve with a correlation. First, candidate regions are extracted by using the characteristics of a parabola that is downward convex, then the candidate region will be verified based on a dissolve modeling error. If a dissolve modeling error on a candidate region is less than a threshold adaptively determined based on the variances between the candidate regions and the target correlation, the candidate region should be a dissolve region with a correlation less than the target correlation. By considering the correlation between neighbor scenes, the proposed method is able to be a semantic scene-change detector. The proposed algorithm was tested on various types of data and its performance proved to be more accurate and reliable when compared with other commonly used method dissolve modeling error. The proposed algorithm consists of two steps. First, the candidate dissolve regions are extracted using the characteristics of the first and second derivate of a variance curve. In the second step, the candidate regions are verified based on a dissolve modeling error. If the dissolve modeling error for a candidate region is less than a threshold defined by a dissolve modeling error with a target correlation, the candidate region is determined as a dissolve region with a lower correlation than the target correlation, which can be given application-dependently by user or can be used as a control factor of video parsing. The proposed algorithm was tested on a variety of data and the performance proved to be more accurate and reliable when compared with other commonly used method.

A study on supervised classification of remote sensing satellite image by bayesian algorithm using average fuzzy intracluster distance

This paper proposes a more effective supervised classification algorithm of remote sensing satellite image that uses the average fuzzy intracluster distance within the Bayesian algorithm. The suggested algorithm establishes the initial cluster centers by selecting training samples from each category. It executes the extended fuzzy c-means which calculates the average fuzzy intracluster distance for each cluster. The membership value is updated by the average intracluster distance and all the pixels are classified. The average intracluster distance is the average value of the distance from each data to its corresponding cluster center, and is proportional to the size and density of the cluster. The Bayesian classification algorithm is performed after obtaining the prior probability calculated by using the information of average intracluster distance of each category. While the data from the interior of the average intracluster distance is classified by fuzzy algorithm, the data from the exterior of intracluster is classified by Bayesian classification algorithm. The testing of the proposed algorithm by applying it to the multispectral remote sensing satellite image resulted in showing more accurate classification than that of the conventional maximum likelihood classification algorithm.

Semi-automatic Video Object Segmentation Method Based on User Assistance and Object Tracking

This paper presents a semi-automatic segmentation method based on user assistance and object tracking. In the method, a user can initially mark objects of interest around the object boundaries. Then the user-guided and selected objects are continuously separated from the unselected areas through time evolution in the image sequences. Experimental results are presented to demonstrate the superiority of the proposed method over automatic methods.

Normalization Domain Watermarking Method Based on Pattern Extraction

This paper presents a new watermarking scheme using the pattern-based image normalization. The proposed method extracts an image-adaptive binary pattern (BP) composed of inner regions, and the geometric moments are computed using the BP rather than the source image itself. This approach avoids any misalignment between the normalized images on both sides of embedding and detection, which is inevitable when the source image itself is used.

New fractal coding scheme using reference images

We describe a simple, fast, and accurate technique for the angular alignment of a polarization-maintaining monomode optical fiber. The method uses simple mechanical supports and is based on the detection of the ellipticity of the light polarization at the fiber output, with the help of a simple rotating polarizer, a photodetector, and an oscilloscope.

A Fast Motion Estimation Using Prediction of Motion Estimation Error

This paper presents a modified MSEA (multi-level successive elimination algorithm) which gives less computational complexity. We predict a motion estimation error using the norms at the already processed levels in the MSEA scheme and then decide on if the following levels should be proceeded using the predicted result. We skip the computation at the following levels where the processing is no longer meaningful. At this point, skipping the processing gives computational gain compared to the conventional MSEA scheme. For the purpose of predicting the norm at each level, we first show the theoretical analysis of the norm at each level and then verify the analysis by experiments. Based on the analysis, a new motion estimation method is proposed and its performance is evaluated.

A fast full search algorithm for multiple reference images

This paper presents a fast full search algorithm for motion estimation. The proposed method is an extended version of the rate constrained successive elimination algorithm (RSEA) for multiple reference frame applications. We will show that motion estimation for the reference images temporally preceding the first reference image can be less intensive in computation compared with that for the first reference image. For computational reduction, we will drive a new condition to lead the smaller number of candidate blocks for the best matched block. Simulation results explain that our method reduces computation complexity although it has the same quality as RSEA.