We propose a digital video scrambling method for multi-level degrading of subjective image quality. Video scrambling methods are used to intentionally degrade image quality to control access for pay-per-view and other video delivery services. Furthermore, digital video scrambling methods that can be used to control the scrambling level have been studied in recent years. However, these conventional methods do not necessarily help in achieving multi-level degradation of subjective image quality. To satisfy this requirement, we introduce the concept of region of interest (ROI). We propose a digital video scrambling method for scrambling partial areas within the ROI and multi-level degrading of subjective image quality using simple scrambling level control while maintaining coding efficiency. Subjective assessment results showed the advantage of the proposed method over conventional methods and its usefulness for various kinds of images.
Digital watermarking technology is often used to embed ID data in analog content (such as an image on printed material), and recently, many algorithms have been proposed for using a camera-equipped cellular phone to detect watermarks in a captured image. One of these, based on information embedding in a frequency space has the advantage of robustness against a partial distortion of the captured image, but it is hampered by a low limit of embedded information quantity. A watermarking method for improving the embedded information quantity by using multiple-valued area of signals is proposed. Multiple-value conversion in the signal area greatly increases the amount of embedded information compared with the conventional binarization algorithm. Experimentation with actual natural images demonstrated the feasibility of the proposed algorithm.
We propose an automatic and precise moving-object extraction method for use in video streams that can also be used for 3-D system applications. The method generates a statistical model for each pixel using several frames, and then uses it to generate trimap images. After manually initializing a frame, unknown regions are automatically determined either background or foreground for the rest of frames. The key technology proposed is an adaptive training scheme, which estimates detection thresholds locally through the algorithm, followed by matting approaches using an iterative process and weighted statistical distance minimization. Experiments demonstrate outperformance of our method for both indoor and outdoor video streams, and also for 3-D modeling and representation.
Large magnification zooming — one of the most important functions in state-of-the-art video cameras — is achieved with a combination of optical-lens and digital zooming. If the magnification ratio is too large, however, optical stabilization mechanisms cannot completely remove unwanted vibrations in a video sequence induced by camera motion. In addition, the zoomed video suffers from blur when simple frame magnification is used for digital zooming. We have developed a novel technique for video stabilization that works in conjunction with super-resolution zooming. The advantage of the proposed technique is that both the stabilization and the super-resolution zooming are performed concurrently in a unified framework. With this framework, the local motions need to be estimated only once for both the stabilization and the zooming. Testing demonstrated that use of the proposed technique results in stable zooming with video resolution much higher than with conventional digital zooming.
We propose a method to improve performance of video coding using an adaptive interpolation filter technique. The adaptive interpolation technique was based on clustering using the k-means method and did not need filter selection information. To improve the performance of the clustering, we introduced autocorrelation coefficients of the pixel values and the directions of the motion vectors as the new features and the Mahalanobis distance as the distance scale in the k-means method. We also used vector quantization to reduce the number of interpolation filter coefficients. In the simulation, our proposed method was implemented in the MPEG-2 based video codec. The simulation results show that the proposed method can reduce the bit rate by up to nearly 7 % compared to conventional adaptive interpolation filter methods.
We describe a lossless digital watermarking technique that uses the characteristics of a number theoretic transform (NTT). Our method generates bit sequences, which restore the original image during the extraction procedure, by using NTT coefficients. These sequences are easy to compress thanks to the coefficients' regularity. The compressed sequences, which are the restoration data, and the side information are embedded in the NTT coefficients of the original image. In the extraction procedure, the original image can be restored by using the restoration data. Because the NTT has no round-off error, this method can completely restore the original image. A simulation experiment was conducted to test the effectiveness of our method.
The use of a novel motion compensated spatio-temporal nonseparable filter (MCSTF) is proposed for scalable video coding. This filter is an alternative to the present motion-compensated temporal filtering (MCTF), a fundamental component for scalable video coding in the next generation. MCTF improve the coding efficiency and provide the temporal scalability. We classify MCSTF as a non-separable sub-sampling version of MCTF, and it provides interlaced pictures as intermediate video sequence by using a spatio-temporal split process. Furthermore, its 1/3-transform structure, which excludes the lifting-update-step, significantly suppresses the PSNR fluctuation that occurs when the existing MCTF technique is used. Our proposed system suppresses PSNR fluctuation with a higher average PSNR than the existing MCTF. Two types of sub-sampling lattices were investigated: the vertical-temporal (VT) quincunx and the face-centered orthorhombic (FCO) ones. Experimental results with entropy coded scalar quantization show the signifaicance of our proposed technique.
Recently, the way to manage working hours has become an issue with the increase in the death rate from overwork. The Labor Standards Law doesn't provide the way to manage working hours, though it makes it obligatory to manage working hours. In general, it is impossible to strictly manage working hours because a worker manually manages his/her own working hours. Therefore, biometric authentication technology using face images is receiving attention. However, the existing technology using face images is unsuitable for time and attendance management because it requires the person to turn his/her face to the camera, taking time and effort. Therefore, we propose a method of managing time and attendance of workers without making the person conscious of the camera. In this research, firstly we extracted a frontal face image by tracking the face in a film. We then identified the individual using the frontal face image. Finally we output time and attendance information for each individual based on the result of identification.
We have proposed a method for reducing the width of panoramic images with high aspect ratios on the basis of the entropy in local color histograms computed for each column in the images. In this paper, we propose a method for enhancing the entropy in the conventional method to further widen and narrow the regions of high and low entropies, respectively. Experimental results show that the proposed method can reduce the width of panoramic images more effectively than the conventional method.
Moving object extraction in video sequences has been studied from various technical viewpoints. We propose a region-merging method for moving object extraction based on robust motion estimation. An over-segmented image is obtained by means of morphological watershed algorithm, and a robust motion estimation method using feature points and an evaluation function composed of motion estimation errors is then applied. Computer simulations demonstrated the merging precision of the proposed method.
In this study, we developed a novel template matching method by establishing two new types of subtemplates for robust recognition of strongly deteriorated signboard images. This method has three main stages: (i) removing occlusion/reflection by using the main peak finding algorithm of uniform color regions; (ii) establishing thinning/thickening and difference sub-templates by using morphological operations; and (iii) adopting template matching criterion based on modified standard deviation. The high performance of our proposed method is shown in the results of an experiment carried out on 200 outdoor images taken under various kinds of bad conditions with occlusion, refection and so on.
The process of deducing personality from fragmentary information about an unknown person is called impression formation. We investigated how the brightness and contrast of facial photographs affect how the impression of personality is formed. We changed the brightness and contrast of facial photographs (male and female average faces) in a stepwise fashion, and assessed their personalities using the semantic differential method. From the results of a factor analysis, three factors were extracted: individual likableness, vitality, and social desirability. We then conducted a two-factor analysis of variance for the subscale scores in each of the three factors. We found that there were significant interactions between brightness and contrast in all three factors for the photographs of female faces. On the other hand, for the photographs of male faces, there was one significant interaction only in the factor of individual likableness. Additionally, the results of simple main effect tests show that brightness affects the impression of individual likableness, contrast affects that of social desirability, and both brightness and contrast affect the impression of vitality.
We propose a method for detecting the duplicated regions in a digital image. Conventional methods treat only a simple case of translation duplication, whereas we treat a complex duplication accompanied with rotation and/or reversal. The proposed method features rotation invariant image features that can be used to estimate rotation angle and judge whether reversal exists (OK?). We can transform an input image by estimating the rotation angle and/or reversal. In an extended image combining the input and transformed images, pixels with similar features are searched to determine the duplicated location. Experimental results show the effectiveness of the proposed method.
An optimal quantization ordering in the feedback quantization scheme for interpolative prediction coding designed to reduce the coding error power is proposed. This method adaptively selects the optimal quantization order to minimize the coding error power from candidate orders for each one dimensional block of 9x1 pixels. These candidate orders can be determined through a numerical analysis of the coding error with respect to all the quantization orders. Simulation results show that the proposed method can effectively reduce the error power and that selecting the optimal order from only four candidate orders can improve the PSNR by about 2 dB, which is just 0.36 dB below the heighest quality image decoded using all 5040 candidate orders. In the case of two-dimensional blocks, although the numerical analysis shows that the coding error power is almost identical among all the possible orders, simulation results demonstrate that selecting the optimal order from four orders in horizontal and vertical directions improves the PSNR by 2.6dB in the case of an 8×8 block.
A wide-band ferrite fin electromagnetic wave absorber panel for a precast curtain wall has been developed for multi-pass interference reduction in urban communication systems. The ferrite fin structure is designed to obtain a wide-band characteristic from the parameter survey of the fin structure. The characteristics are measured using mock-ups of the precast curtain wall with a time-domain method. Results indicate that the proposed absorber panel suppresses reflection < -14 dB in the frequency band 88.6 - 916.6 MHz for the incident angle of 30 degrees and 80.0 - 987.4 MHz for that of 60 degrees. The absorber panel can be applied to the TV ghost problem and other communication systems in both VHF and UHF bands.
In low-voltage CMOS digital circuits, the threshold voltage variation causes significant circuit performance fluctuation: we therefore propose on-chip process compensation techniques for such circuits. We used voltage reference circuits, that can monitor process variations. We confirmed the operation of the circuit by using a SPICE simulation with a set of 0.35-μm standard CMOS parameters, and we performed Monte Carlo simulations assuming process spread and device mismatch in all MOSFETs. SPICE simulation demonstrated that the process variations of digital circuits were improved by 45% after applying the proposed architecture. These techniques will be useful for the on-chip process compensation of low-voltage digital circuits.
We propose an effective arrangement of packet schedulers (WFQ and FIFO) that can enhance the QoE of an audio-video streaming transmission over IP networks (typified by IPTV). Previously, the authors examined the relationship between the arrangement of WFQ and application-level QoS and observed that WFQ for all or both edge nodes leads to a high application-level QoS. Now, we have assessed the arrangement in terms of the QoE in addition to application-level QoS. The results show that WFQ on the receiver-side edge in the network satisfactorily enhances the QoE as well as the WFQ on both edges and the WFQ on all nodes. This implies that QoS control in the access network dominates QoE.
The necessity of ultrasonic diagnosis tools increases every year. We propose an automatic endocardium tracing method that applies active contour models(ACM) in two stages. In the first stage, an endocardium contour is detected using the result of an edge extraction based on the separability of image features. In the second stage, the endocardium contour is extracted using shape correction processing. "Mitral valve processing" not only detects the position of the mitral valve at the end-diastolic period but also corrects the detected contour after the first stage of ACM. The result of experiments using one healthy and three diseased cases demonstrate the effectiveness of the proposed method.