Video content management methods (VCMMs) have been studied in order to reduce the delivery loads in cache-based video content distribution systems (VCDSs). A VCDS using multiple-quality require a VCMM with hierarchical data. However, due to the occurrence of priority estimation errors (PEEs), there is a problem of reduction of cache efficiency in the practical variation models, e.g., increase/decrease of access number, the presence or absence of updating a video content and the change of the video content popularity. Then, the practical temporal-variation-models (PTVM) are firstly defined, and this paper proposes a multiple-Quality management method using an adaptive priority estimation method (APEM), which decreases PEEs in the practical models. After the occurrence states of PEEs in the estimation method (EM) are considered, an APEM for those states is shown. This study prepares three approximation functions based on a least-square method as EMs, i.e., a linear approximation function, a quadratic approximation function and an exponential approximation function. The proposed method (PM) aims at the avoidance of reduction in cache efficiency due to PEEs for the increase/decrease states of the estimation values. By the evaluation experiments, comparing PM with the conventional methods (CM) which independently use each function, it was revealed that PM could stably control the reduction of cache efficiency. By the experimental results, in PTVM, it was also found that there was the time zone in which the efficiency decreased in CM and the efficiency in PM could be improved 18.49% greater than that in CM in the cache hit rate within the time zone. In this case, PM achieved 5.08×10⁻¹% improvement for CM in average cache hit rate.

Eye trackers have been used in various fields such as eye movement-related brain activity analyses, visual attention analyses and gaze input interfaces.Recently,these trackers have been improved in both terms of accuracy and portability. In this study, we propose a new eye tracking method which enables to estimate gaze direction from faces in images or movies without using the conventional eye trackers, and evaluate the proposed method based on accuracies of point-of-regards (POR) to measure its capability as gaze input interfaces. The proposed method contains four steps: fitting a 3D model to the targeted facial image, extracting and geometrically correcting the regions of eyes using head-pose information, fitting a circle to each iris to localize its center, and compensating the POR corresponding to the head-pose changes. Our experiment’s result on 11 subjects shows that the average error between each fixation target and the POR is about 2 and 3.5 degrees for ±5 and ±5～±20 degrees head movement, respectively.

In this paper, new pencil art generation method which focuses on pencil line expression is proposed. In proposed method, outline of pencil art is generated by Pencil stroke map, and its shading is generated by LIC (Line Integral Convolution) method. The both method have features such as they are excel at expressing line of pencil art. Pencil stroke map can generate high quality outline of pencil art by overwriting with short pencil line, but if the length of lines are too long for size of graphic, there is a problem that resulting outline lose its shape. On this problem, this study improve that by classifying graphics of input image by size by morphological operation and generate resulting outline with plural length of pencil line. And, hatching texture which is basic technique of pencil art can be generated by LIC method based on gray level of input image, Moreover, this study show that the hatching can be cross-hatching by overwriting themselves, and blending can be generated by applying smoothing filter. Proposed method generate high quality pencil art by combining those two method. In experiment, Effectiveness of proposed method is shown by comparing with multiple conventional methods.

It has been known that the feeling on the texture surface can be improved by adding Gaussian noise to a given original image. It is laborious to determine the appropriate noise intensity for the texture feeling by subjective evaluation experiments with many raters. In this paper, the objective estimation method of the appropriate noise intensity to reconstruct the optimum memory texture feeling is proposed based on fractal dimension using a set of noisy images. It is seen in the experimental results using the paired comparison method that the estimated noise intensity by our method can improve the memory texture feeling compared with other noise intensities.

Recent years have seen the use of RGB-D sensors such as Kinect in three.dimensional environmental map construction with robots. RGB-D sensors are advantageous as they are inexpensive and easy to use. However, these sensors cannot perform high.precision measurements. Consequently, the depth information obtained from them may be distorted. The authors of this paper investigated calibration techniques for RGB-D sensors with the aim of autonomously constructing high.precision three.dimensional environmental maps with indoor mobile robots. The findings showed that the RGB-D sensor’s depth measurement errors vary for each sensor and image pixel. Furthermore, by using two types of linear functions for long and short distances, it was possible to correct the depth. However, two drawbacks of previous study were not resolved leading to measurement errors. The correction results for depth data with large distortions had low accuracy, and the boundaries of the correction formulae for short and long distances were discontinuous. In this paper, the correction method of the RGB-D sensor was improved. In addition, a comparative analysis was performed for the accuracy of environmental maps after corrections.

Marching is an activity of wind orchestra, where players move according to a preconcerted route and change their formation. In this paper, we introduce a system for visually analyzing the accuracy of player’s route from a marching movie recorded through a commercially available video camera. This system can identify a player captured in the movie using an algorithm for object tracking and show differences between the player’s present route and the originally designed one. Proposed user interface allows users to select a player to analyze his / her trajectory and play the movie with the visualized player’s route. Even when the tracked player is hidden behind another player or an instrument, the system can robustly predict the position of the player using the past movement vectors. The system makes a player notice his / her points of improvement, and thus the completeness of a formation in marching is expected to be enhanced.