Yunsang Han

Pixel Correlation-based Impulse Noise Reduction

Median filtering based on switching is known that one of the most effective algorithms for reducing impulse noise. In this paper, we introduce a Pixel Correlation based Impulse Noise Reduction. The proposed algorithm consists of two main steps, detection and correction. In the first step, noise detection is performed using a scalable detection mask and morphological operation. In the second step, a corrupted pixel is corrected using the correlation between the uncorrupted pixels in the mask. The experimental results showed that the proposed method can reduce more impulse noise and preserve more edge information than state-of-the-art standard median filter and switching median filters (adaptive median filter and iterative adaptive switching median filter) by an average of 2.75 and 0.37 dB, respectively.

Parameter estimation-based single image super resolution

In this paper, we introduce a parameter estimation-based single image super resolution technique. The basic idea of the proposed method is to use the property of unknown high resolution image inferred by relations of its lower resolution images. The proposed algorithm is consists of 3 main phases: 1) in the first step, an error model between an input and its lower resolution images is constructed for inferring the property of unknown high resolution image; 2) In the second step, global enhancement using estimated signal complexity and its strength is performed. 3) Lastly, independent enhancement in artifact candidate regions is performed in estimated artifact candidate regions with artifact removal using property of morphological operation. The experimental results show the efficiency of the proposed algorithm compared to a state-of-the-art method. Besides, the proposed method is not only much faster than the compared method but also able to implement in hard ware such as digital TVs and smart phones. Therefore, we believe that the proposed method can be a useful tool for super resolution-related consumer electronics fields.

New and efficient dynamic capacitance compensation (DCC) system to reduce data size and artifacts for 120Hz Full-HD LCD panel

Although the overdrive technique is one of the most effective techniques for reducing motion blur artifact in liquid crystal displays (LCD), it requires considerable frame memory (more than one frame). Artifacts can be reduced significantly when more frames are used for overdrive technology. This paper introduces a new and efficient dynamic capacitance compensation (DCC) system to reduce motion blurring as well as other artifacts, such as overshoot and rising bounce using 3 frames. The main idea of the proposed system is to use 3 frames using motion estimation for the overdrive of a 120Hz Full-HD LCD panel. To evaluate the performance of the proposed method, the quantitative measures were defined in terms of the peak signal-to-noise (PSNR) and data reduction ratio. The experimental results showed that proposed algorithm could reduce data size significantly (~90%), and the resulting images had sufficient quality (average 34.68dB) for overdrive in the LCD panel. Therefore, the proposed approach can be a useful tool for reducing the frame data size of the LCD overdrive technique as well as artifacts, such as motion blurring, overshoot and rising bounce.

Non-parametric single image super resolution

In this paper, we introduce a single image super resolution based on non-parametric local information. The basic idea of the proposed method is to use a property, which is inferred by relations between input and its lower resolution images, of an unknown high resolution image. The experimental results show the efficiency of the proposed algorithm compared to a state-of-the-art method. Besides, the proposed method is not only much faster than the compared method but also able to implement in hardware such as digital TVs and smart phones. Therefore, we believe that the proposed method can be a useful tool for super resolution-related fields.

Structure based noise reduction using LPA-ICI

In this paper, we introduce a structure-based noise reduction algorithm using LPA-ICI technique. The proposed scheme consists of two main steps: In the first step, we obtained a scale map using LPA-ICI technique to determine the proper criterions, such as mask size and search range, during the filtering step; In the second step, we perform the filtering which considers the local structure between a current block and neighboring blocks for preserving local information. The experimental results showed that the proposed method could significantly reduce noise and preserve the detail and edge information. It is believed that it can be used as a good tool for noise reduction in the field of consumer electronics.