Suk-Ju Kang

Perceptual Quality-Aware Power Reduction Technique for Organic Light Emitting Diodes

In this paper, a perceptual quality-aware power reduction algorithm is proposed. Structural similarity (SSIM) is a widely used metric to evaluate perceptual image quality between original and processed images in a display. The proposed algorithm linearly changes the curve for the gray-level mapping between the input and output considering the target SSIM value. It then modifies the SSIM equation based on the modified curve and extracts the solution, which is the slope of the curve, from the equation. Finally, the algorithm optimally controls image quality by changing the slope of the gray-level mapping curve. Simulation results demonstrate that the proposed algorithm preserves image quality, regardless of image characteristics. In particular, the average SSIM and peak signal-to-noise-ratio values of the proposed algorithm were up to 0.021 and 11.203 dB higher than those of the benchmark algorithm. In addition, the average computation time of the proposed algorithm was up to 0.619 μs per pixel lower than that of the benchmark algorithm.

Photosensor-Based Latency Measurement System for Head-Mounted Displays

In this paper, a photosensor-based latency measurement system for head-mounted displays (HMDs) is proposed. The motion-to-photon latency is the greatest reason for motion sickness and dizziness felt by users when wearing an HMD system. Therefore, a measurement system is required to accurately measure and analyze the latency to reduce these problems. The existing measurement system does not consider the actual physical movement in humans, and its accuracy is also very low. However, the proposed system considers the physical head movement and is highly accurate. Specifically, it consists of a head position model-based rotary platform, pixel luminance change detector, and signal analysis and calculation modules. Using these modules, the proposed system can exactly measure the latency, which is the time difference between the physical movement for a user and the luminance change of an output image. In the experiment using a commercial HMD, the latency was measured to be up to 47.05 ms. In addition, the measured latency increased up to 381.17 ms when increasing the rendering workload in the HMD.

Real-time stereo matching using extended binary weighted aggregation

This paper presents an accurate real-time stereo matching method, which is based on the extended binary weighted aggregation. The accuracy of the proposed stereo matching method was significantly enhanced by extending its binary weighted aggregation so that remote connections of support regions can be allowed for aggregation. The extended binary weighted aggregation is based on the following two new ideas. First, the extended binary weighted aggregation connects distant regions over color boundaries, making them one large support region for a given pixel. This approach induces more aggregation targets, and, thus, makes the aggregation step more robust. Second, it excludes cost outliers in the support region to prevent them from being propagated during the aggregation step, making a quality support region. With the extended binary weighted aggregation, the proposed stereo matching method obtains more accurate disparity maps than existing stereo matching methods using binary weighted aggregation methods, while maintaining the speed advantage of binary weighted aggregation. Experimental results illustrated that the proposed stereo matching method outperformed all existing real-time stereo matching methods in terms of accuracy, providing the average bad pixel rate of 5.12%, for the Middlebury stereo test images. The proposed stereo matching method was implemented on a CUDA platform with a high-end GPU. The implemented system operated at up to 300 fps for the stereo images with 320 × 240 pixel resolution and a disparity range of 32 pixels.

Image Segmentation Using Linked Mean-Shift Vectors and Global/Local Attributes

This paper proposes novel noniterative mean-shift-based image segmentation that uses global and local attributes. The existing mean-shift-based methods use a fixed range bandwidth, and hence their accuracy is dependent on the range spectrum of an image. To resolve this dependency, this paper proposes to modify the range kernel in the mean-shift process to be anisotropic. The modification is conducted using a global attribute defined as the range covariance matrix of the image. Further, to alleviate oversegmentation, the proposed method merges the segments having similar local attributes more aggressively than other segments. The local attribute for each segment is defined as the sum of the variances of the chromatic components. Finally, to expedite the processing, the proposed method uses a region adjacency graph (RAG) for the merging process, thus differing from the existing linked mean-shift-based methods. In the experiments on the Berkeley segmentation data set, the use of the global and local attributes improved segmentation accuracy; the proposed method outperformed the state-of-the-art linked mean-shift-based method by showing an improvement of 2.15%, 3.16%, 3.32%, and 1.90% in probability rand index, segmentation covering, variation of information, and F-measure, respectively. Further, compared with the benchmark method, which uses the dilating and merging scheme, the proposed method improved the speed of the merging process 42 times by applying the RAG.

Prediction-based latency compensation technique for head mounted display

In this paper, we propose a prediction-based latency compensation system for head mounted display. Specifically, the proposed system uses a linear extrapolation of head orientations for prediction based on biological data of body. The experimental results show that the proposed system compensates a latency up to 53 milliseconds with 1.083 degrees of a minimum average error.

Adaptive contrast enhancement using edge-based lighting condition estimation

Abstract This paper proposes a new approach to image contrast enhancement that improves the perceptual visual quality by considering the lighting condition and minimizing the structural distortion to a tolerable level. The proposed method consists of the following two major steps: lighting condition estimation and contrast enhancement processes. In the first step, the proposed method estimates the lighting condition by calculating the dynamic range along the edges of the image. In the second step, the method adaptively adjusts the luminance by considering both the estimated lighting condition and the order of luminance levels in order to improve the perceptual visual quality. In addition, the method properly reduces the structural distortion. Experimental results show that the proposed method improved the perceptual visual quality of various images by increasing the average structural fidelity, enhancement performance measure, entropy, and tone-mapped image quality index by up to 11%, 133%, 16%, and 11%, respectively, compared to the benchmark methods.

Real-time temporal quality compensation technique for head mounted displays

An asynchronous time warp (ATW) is used to reduce a motion-to-photon latency in head mounted displays (HMDs). To implement the ATW, a graphics processing unit (GPU) must allow to perform the time warp during the image rendering. However, typical GPUs do not support the preemption optimized for the ATW. Even though the preemption is performed in the GPUs, visual artifcat like a judder can occur. Therefore, high priced GPUs with high performance optimized for the ATW must be used to enhance the temporal image quality. In this paper, a new field-prgrammable gate array (FPGA) based HMD system is proposed to perform the temporal quality compensation without using ATW. The proposed system can enhance the perceived image quality in the HMD even when using the typical GPUs with low performance. We develop the hardware system on the low priced FPGA handling output images with the 1440X2560 pixel resolution. Experimental results show that the average PSNR of our system is 30.15dB for the temporal image quality and the total execution time of this system is 1.48 ms. Thus, our system can be implmented in real-time with the improved temproal image quality.

Dual-dissimilarity measure-based statistical video cut detection

Video cut detection is an essential process of temporal continuity-based video applications such as video segmentation, video retargeting, and frame rate up-conversion. The performance of these applications highly depends on the performance of cut detection. This paper proposes an effective and low-complexity approach for detecting video cuts. The proposed method uses two simple dissimilarity measures for video cut detection: inter-frame luminance variation and temporal variation of inter-frame variations over several frames. The first is used to detect abrupt changes, and the second is used to reduce the influence of disturbances, e.g., object or camera motion. The proposed method is comprised of the following three steps. First, it computes the two dissimilarity measures. Then, it combines them using Bayesian estimation and linear regression. Finally, it decides on the possibility of cuts using the combined dissimilarity measure. Experimental results show that the average F1 score of the proposed method was up to 0.252 (37.0%) higher than those of the benchmark methods. Moreover, the algorithmic simplicity of the proposed method reduced the average computation time per pixel by up to 99.8%, when compared with state-of-the-art methods. Thus, the proposed method is superior to existing methods in terms of computational complexity and detection accuracy.

Backlight dimming based on saliency map acquired by visual attention analysis

Abstract Displays have been used in various applications from mobile phones to tablets, and the low power consumption is one of their most important factors. Backlight dimming is the most promising technique to achieve this because it significantly reduces the display power by controlling only the transmittance of liquid crystal. This paper proposes a new backlight dimming algorithm using visual attention analysis. Conventional algorithms have a serious saturation error in some images when performing backlight dimming, thereby degrading image quality. In contrast, the proposed algorithm analyzes image characteristics based on the saliency map, which considers human visual attention. It then truncates the meaningless information of the saliency map using an adaptive saliency level selection approach and calculates the maximum amount of saturation error that humans will not perceive. In addition, the proposed algorithm defines the objective function and computes the optimal starting gray level in that function to calculate the saturation error. Simulation results show that the proposed algorithm using the adaptive saliency level selection approach performs best. In addition, the average peak signal-to-noise ratio of the proposed algorithm was up to 3.762 dB higher than that of the conventional algorithm while slightly increasing the power consumption.

Motion vector smoothing of boundary of moving object for frame rate up-conversion

This paper presents a motion vector smoothing method to increase the accuracy of the motion vector in the boundary of moving object. Existing motion vector smoothing methods cannot generate the accurate motion vectors in the boundary of moving object and it induces the block artifacts. To reduce these block artifacts, the proposed method detects the object boundary and extracts the moving region using the temporal information. Then, the boundary of moving object is extracted using the object boundary and the moving region information. Finally, the motion vector smoothing is performed to correct the falsely detected motion vector in the boundary of moving object. Experimental results demonstrate that the proposed method successfully increased the peak signal-to-noise ratio and the structural similarity index by up to 4.12 dB and 0.0626, respectively, compared to the benchmark methods.