발광다이오드 백라이트를 사용한 저전력, 고화질 액정표시장치를 위한 로컬 디밍 알고리즘

Abstract

액정표시장치는 얇은 두께, 가벼운 무게, 높은 해상도와 같은 장점들 때문에 다양한 디스플레이 제품에서 사용된다. 그러나 입력영상의 특성에 상관없는 일정한 백라이트 밝기로 인하여 전력소모가 상당히 증가되고, 최하 명암도에서 액정의 빛누설로 인하여 출력영상의 명암비가 낮아진다. 이러한 문제점을 해결하기 위해, 로컬 디밍 알고리즘이 연구되어 왔다. 로컬 디밍 알고리즘에서는 백라이트가 여러 개의 로컬 블록으로 나뉘고, 입력 영상의 특성에 따라 로컬 블록의 밝기가 각각 조절된다. 입력 영상의 어두운 부분에 있는 로컬 블록의 밝기를 감소시켜서 백라이트의 전력소모를 감소시키고 명암비를 증가시킨다. 본 논문에서는 다양한 종류의 발광다이오드 백라이트를 사용하는 액정표시장치에 적용 가능한 로컬 디밍 알고리즘을 제안하였다. 또한 제안된 로컬 디밍 알고리즘의 성능은 모의실험과 실험의 결과를 통해서 검증하였다. 첫 번째로, 본 논문에서는 백라이트의 전력소모를 줄이고 명암비를 증가시키기 위해서 백색 발광다이오드 백라이트를 사용하는 액정 표시장치에 적용 가능한 그레이 로컬 디밍 알고리즘을 제안하였다. 제안된 알고리즘은 두 개의 파라미터간의 관계를 이용하여 간단한 구동과 다양한 영상에 적합한 결과를 달성하였다. 백라이트의 밝기는 평균 밝기에 최대 밝기와 평균 밝기의 차이를 더하는 방식으로 결정된다. 평균 밝기는 입력 영상의 휘도 정보를 반영하고, 최대 밝기와 평균 밝기의 차이는 입력 영상의 대비를 보상하기 위해 사용된다. 모의실험 결과, 평균 전력감소율은 50.2%, 평균 최대 신호 대비 잡음비는 40.4 dB, 구조 유사성은 0.995이다. 제안된 그레이 로컬 디밍 알고리즘은 백색 발광다이오드 백라이트를 사용한 7인치 박막 트랜지스터 액정표시장치에서 실연되었다. 측정된 평균 전력감소율은 21%이다. 두 번째로, 컬러 발광다이오드를 사용한 액정표시장치의 영상왜곡과 전력소모를 줄이기 위해서 컬러 로컬 디밍 알고리즘을 제안하였다. 제안된 알고리즘은 RGB 발광다이오드 백라이트를 사용하는 55인치 액정표시장치에서 실연되었다. 제안된 알고리즘에서는 안정적인 화질은 얻기 위해서 디밍 영상의 영상왜곡을 계산하여 로컬 블록의 밝기를 결정하게 된다. 또한, 디밍 영상의 색왜곡을 줄이기 위해서 색좌표 상의 차이를 제거하도록 영상데이터를 조정한다. 테스트 동영상으로 모의 실험한 결과, 평균 전력감소율은 66.9%, 평균 최대 신호 대 잡음비는 43.7 dB, 최대 신호 대 잡음비의 평균 표준편차는 3.6 dB이다. 제안된 알고리즘을 적용한 평균 색편차는 2×2, 4×2, 4×4, 8×4의 화이트박스에서 각각 9.63, 5.03, 5.55, 4.15이다. 측정된 평균 전력소모는 67.3%이다. 마지막으로, 엣지 타입 발광다이오드 백라이트를 사용하는 액정표시 장치에 적용 가능한 로컬 디밍 알고리즘을 제안하였다. 백라이트 가장자리에 직교방향으로 위치한 발광다이오드의 행렬구동을 통해서 제안된 로컬 디밍 알고리즘을 구현하였다. 발광다이오드의 디밍 비율은 로컬 블록의 목표 밝기에 따라 결정되고, 영상왜곡을 줄이기 위해 조절된다. 모의실험 결과, 평균 전력감소율은 26.3%, 평균 최대 신호 대비 잡음비는 44.7 dB, 구조 유사성은 0.999이다. | Liquid crystal displays (LCDs) are used in various display products due to their desirable qualities of thinness, light weight, and high resolution. However, regardless of input image characteristics, their power consumption is appreciably increased by constant backlight luminance, and the contrast ratio of the displayed image is lowered by liquid crystals’ light leakage at lowest gray level. To solve these problems, local dimming algorithms have been researched. In the local dimming algorithm, the backlight is divided into several local blocks, and their luminance is controlled separately according to the input image characteristics. The power consumption of the backlight can be reduced and the contrast ratio can be improved by decreasing the backlight luminance of local blocks in dark parts of the input image. This dissertation proposes local dimming algorithms for LCDs using various types of light emitting diode (LED) backlights. The performances of the proposed local dimming algorithms are evaluated by simulation and experiment results. Firstly, this dissertation proposes a gray local dimming algorithm for LCDs using white LED backlight to reduce power consumption of backlight and improve contrast ratio. The proposed algorithm is a simple operation and it is well adaptive for various images using relationship of two parameters. The backlight luminance is determined by summing up average luminance and difference between average and maximum luminance. The average luminance is used to reflect brightness information of input image, and the difference between average and maximum luminance is used to compensate contrast of input image. Simulation results show that the average power saving ratio, the average peak signal to noise ratio (PSNR), and the average structural similarity (SSIM) for test images are 50.2%, 40.4 dB, and 0.995, respectively. The proposed gray local dimming algorithm is demonstrated on a 7-inch TFT-LCD with LED backlight. The measured average power saving ratio is 21%. Secondly, a color local dimming algorithm is proposed for LCDs using color LED backlight to reduce image distortion and power consumption. The proposed algorithm is demonstrated using a 55-inch LCD panel with an RGB LED backlight system. In the proposed algorithm, the backlight luminance of local block is determined based on the calculated image distortion of the dimming image to achieve steady image quality. The image data are modified to reduce color distortion by removing chromatic difference. Simulation results show that the average power saving ratio, the average PSNR, and the standard deviation of PSNRs for the test video are 66.9%, 43.7 dB, and 3.6 dB, respectively. The average color differences for 2×2, 4×2, 4×4, and 8×4 white boxes are 9.63, 5.03, 5.55, and 4.15, respectively. The measured average power saving ratio is 67.3%. Finally, a local dimming algorithm is proposed to reduce power consumption of LCDs using edge-type LED backlight. The proposed local dimming algorithm is implemented by matrix driving of orthogonally located LEDs at the edge of the backlight. The dimming rate of LED string is determined according to target luminance of local blocks and modified to reduce image distortion. Simulation results show that the average power saving ratio, the average PSNR, and the average SSIM for test images are 26.3%, 44.7 dB, and 0.999, respectively.; Liquid crystal displays (LCDs) are used in various display products due to their desirable qualities of thinness, light weight, and high resolution. However, regardless of input image characteristics, their power consumption is appreciably increased by constant backlight luminance, and the contrast ratio of the displayed image is lowered by liquid crystals’ light leakage at lowest gray level. To solve these problems, local dimming algorithms have been researched. In the local dimming algorithm, the backlight is divided into several local blocks, and their luminance is controlled separately according to the input image characteristics. The power consumption of the backlight can be reduced and the contrast ratio can be improved by decreasing the backlight luminance of local blocks in dark parts of the input image. This dissertation proposes local dimming algorithms for LCDs using various types of light emitting diode (LED) backlights. The performances of the proposed local dimming algorithms are evaluated by simulation and experiment results. Firstly, this dissertation proposes a gray local dimming algorithm for LCDs using white LED backlight to reduce power consumption of backlight and improve contrast ratio. The proposed algorithm is a simple operation and it is well adaptive for various images using relationship of two parameters. The backlight luminance is determined by summing up average luminance and difference between average and maximum luminance. The average luminance is used to reflect brightness information of input image, and the difference between average and maximum luminance is used to compensate contrast of input image. Simulation results show that the average power saving ratio, the average peak signal to noise ratio (PSNR), and the average structural similarity (SSIM) for test images are 50.2%, 40.4 dB, and 0.995, respectively. The proposed gray local dimming algorithm is demonstrated on a 7-inch TFT-LCD with LED backlight. The measured average power saving ratio is 21%. Secondly, a color local dimming algorithm is proposed for LCDs using color LED backlight to reduce image distortion and power consumption. The proposed algorithm is demonstrated using a 55-inch LCD panel with an RGB LED backlight system. In the proposed algorithm, the backlight luminance of local block is determined based on the calculated image distortion of the dimming image to achieve steady image quality. The image data are modified to reduce color distortion by removing chromatic difference. Simulation results show that the average power saving ratio, the average PSNR, and the standard deviation of PSNRs for the test video are 66.9%, 43.7 dB, and 3.6 dB, respectively. The average color differences for 2×2, 4×2, 4×4, and 8×4 white boxes are 9.63, 5.03, 5.55, and 4.15, respectively. The measured average power saving ratio is 67.3%. Finally, a local dimming algorithm is proposed to reduce power consumption of LCDs using edge-type LED backlight. The proposed local dimming algorithm is implemented by matrix driving of orthogonally located LEDs at the edge of the backlight. The dimming rate of LED string is determined according to target luminance of local blocks and modified to reduce image distortion. Simulation results show that the average power saving ratio, the average PSNR, and the average SSIM for test images are 26.3%, 44.7 dB, and 0.999, respectively.