Samuel Moon-Ho Song

Processing of partial video data for detection of wipes

With the currently existing shot change detection algorithms, abrupt changes are detected fairly well. It is thus more challenging to detect gradual changes, including fades, dissolves, and wipes, as these are often missed or falsely detected. In this paper, we focus on the detection of wipes. The proposed algorithm begins by processing the visual rhythm, a portion of the DC image sequence. It is a single image, a sub-sampled version of a full video, in which the sampling is performed in a predetermined and systematic fashion. The visual rhythm contains distinctive patterns or visual features for many different types of video effects. The different video effects manifest themselves differently on the visual rhythm. In particular, wipes appear as curves, which run from the top to the bottom of the visual rhythm. Thus, using the visual rhythm, it becomes possible to automatically detect wipes, simply by determining various lines and curves on the visual rhythm.

Fast algorithm for detection of camera motion

Most image analysis/understanding applications require accurate computation of camera motion parameters. However, in multimedia applications, particularly in video parsing, the exact camera motion parameters such as the panning and/or zooming rates are not needed. The detection--i.e., a binary decision--of camera motion is all that is required to avoid declaring a false scene change. As camera motions can induce false scene changes for video parsing algorithms, we propose a fast algorithm to detect such camera motions: camera zoom and pan. As the algorithm is only expected produce a binary decision, without the exact panning/zooming rates, the proposed algorithm runs on a reduced data set, namely the projection data. The algorithm begins with a central portion of the image and computes the projection data (or the line integrals along the x- or y-axis) to turn the 2D image data into a 1D data. This projected 1D data is further processed via correlation processing to detect camera zoom and pan. Working with projection data saves processing time by an order of magnitude, since for instance, a 2D correlation takes N2 multiplies per point, however a 1D correlation takes N multiplies per point. The efficacy of the proposed algorithm is tested for a number of image sequences and the algorithm is shown to be successful in detecting camera motions. The proposed algorithm is expected to be beneficial for video parsers working with Motion-JPEG data stream where motion vectors are not available.

Motion estimation by Fermat number transform

A new block-matching algorithm, based on the Fermat Number Transform (FNT), is presented. It declares the most correlated-block as the best matching block, as opposed to declaring the block with the least sum of differences between the blocks. The proposed number theoretic approach significantly reduces the computational complexity for the estimation process.

Elimination of blocky artifacts in predictive video coding

In a number of image compression standards, the motion vectors are generally obtained by the well-known block matching algorithm (BMA), the error image along with the computed motion vectors are encoded. Unfortunately, this widely used approach generates artificial block boundary discontinuities, called blocky artifacts, between the blocks. Since the blocky artifacts are caused by synthesizing the predicted frame using one constant motion vector per block, we propose an algorithm that interpolates the motion vectors before the construction of the predicted image. Naturally, using spatially smooth motion vectors completely eliminates the blocky artifact. However, we can no longer use the motion vectors as provided by the BMA. The optimum motion vectors must minimize the norm of the error image. The proposed algorithm computes the optimum motion vectors, with the interpolation process built into the algorithm. To obtain spatially smooth motion vectors, we use a band-limited interpolation, and thus, we refer to our algorithm as the band-limited motion compensation (BLMC). Our simulations indicate that the BLMC completely eliminates the blocky artifacts, as expected, and in addition provides higher peak-signal-to-noise-ratio in comparison to the traditional BMA based motion compensation (BMC) as well as the overlapped BMC.

Morphological approach to scene change detection and digital video storage and retrieval

With the abstraction of digital video, as the corresponding binary video, a process which, upon subjective experimentation seems to preserve the intelligibility of video content, we can pursue a precise and analytic approach to digital video storage and retrieval algorithm design based upon geometrical and morphological intuition. The foremost and tangible general benefit of such abstraction, however, is the immediate reduction of both data and computational complexities, involved in implementing various algorithms and databases. The general paradigm presented may be utilized to address all issues pertaining to video library construction, including visualization, optimum feedback query generation, and object recognition. However, the primary focus of attention in this paper pertains to detection of fast and gradual scene changes, such as dissolves, fades, and various special effects, such as wipes. Upon simulation, we observed that we can achieve performances comparable to those of others with drastic reductions in both storage and computational complexities. The conversion from grayscale to binary videos can be performed directly (with minimal additional computation) in the compressed domain by thresholding on the DCT DC coefficients themselves, or by using the contour information attached to MPEG4 formats. The algorithms presented herein are ideally suited for performing fast (on-the-fly) determinations of scene change, object recognition, and/or tracking, as well as other, more intelligent, tasks, traditionally requiring heavy demand of computational and/or storage complexities. The fast determinations may then be used on their own merit , or can be used in conjunction/complement with other higher-layer information in the future.

Band-limited interpolation for maximum intensity projection images

X-ray computed tomography (CT) scanners, due to its inherent scanning geometry, provide images with non-isotropic voxels. The typical CT scanner generates axial slices with the thickness on the order of a few millimeters with sub- millimeter pixels. The multi-slice images obtained with such protocol must be then interpolated across the slices for an effective and realistic 3-D visualization of the patient anatomy. In this paper we focus on the effects of slice interpolation for maximum intensity projection (MIP) images with the projection direction orthogonal to the z-axis, for instance, for the generation of coronal or sagittal views. Linear interpolations, although simple, due to the inherent noise in the data, generate MIP images with noise whose variance vary quadratically along the z-axis. As such, the MIP images will often suffer from horizontal streaking artifacts, exactly at the position of the original slices. To combat this situation we have developed a different interpolation technique based on a digital finite impulse response (FIR) filter. It is shown that this band-limited interpolation based on the FIR filter will flatten the change in the noise variance along the z-axis, the net effect being either the elimination or a reduction of the horizontal streaking artifact.

Morphological approach to smoothing

In this paper, we present some fundamental theoretical results pertaining to the question of how many randomly selected labelled example points it takes to reconstruct a set in euclidean space. Drawing on results and concepts from mathematical morphology and learnability theory, we pursue a set-theoretic approach and demonstrate some provable performances pertaining to euclidean-set-reconstruction from stochastic samples. In particular, we demonstrate a stochastic version of the Nyquist Sampling Theorem - that, under weak assumptions on the situation under consideration, the number of randomly-drawn example points needed to reconstruct the target set is at most polynomial in the performance parameters and also the complexity of the target set as loosely captured by size, dimension and surface-area. Utilizing only rigorous techniques, we can similarly establish many significant attributes - such as those relating to robustness, cumulativeness and ease-of- implementation - pertaining to smoothing over labelled example points. In this paper, we formulate and demonstrate a certain fundamental well-behaving aspect of smoothing.

Optimization of radar pulse compression processing

We propose an optimal radar pulse compression technique and evaluate its performance in the presence of Doppler shift. The traditional pulse compression using Barker code increases the signal strength by transmitting a Barker coded long pulse. The received signal is then processed by an appropriate correlation processing. This Barker code radar pulse compression enhances the detection sensitivity while maintaining the range resolution of a single chip of the Barker coded long pulse. But unfortunately, the technique suffers from the addition of range sidelobes which sometimes will mask weak targets in the vicinity of larger targets. Our proposed optimal algorithm completely eliminates the sidelobes at the cost of additional processing.