A novel method of combining pixel density map and SLIC for low-power display

OLED displays have different power consumption depending on R, G, and B pixel values. Therefore, if an image is segmented according to saliency and then divided according to color using a super pixel algorithm, low power can be achieved while maintaining human visual satisfaction However, if the image is segmented using saliency and then the segmented image is segmented using the super pixel algorithm, simple linear iterative clustering (SLIC), the pixels that do not have a color value because the segmented image has a different saliency level are also segmented by the super pixel algorithm. So the ability to divide color is poor at segmented image. This paper excludes pixels that do not have color values from the segmentation process when dividing an image including pixels that do not have color values by saliency criteria into super pixels. In addition, by allocating the first search position not evenly in the entire image, but focusing on the pixels with color values, the performance of the super pixel that divides the image according to color in the image divided based on saliency was improved. In terms of low power, the proposed method has similar power savings of about 38% to that of the FDM-oriented SLIC method, but the SSIM, which is structurally similar to the original image, has shown higher.