Search

Medical reports show that as humans age, they experience color perception loss. It is caused by the yellowing occurring on the human lens and the reduction of the human pupil size. These occurrences change the spectral characteristics and reduce the amount of light at the retina. Thus, the images perceived become darker and yellowish as we grow older. Due to this, it may affect the accuracy of the information obtained from color images on websites and smartphone applications. To understand the difficulties experienced by the elderly, researchers have proposed methods to model their color perception. In this paper, we conduct a study on three existing methods that simulate elderly color perception. The study aims to understand their theoretical foundations and mathematical formulation. Also, we study the age-related change to the chromaticity and colors of the simulated images from these methods. Moreover, we also compute the color difference between the original and simulated images. [ABSTRACT FROM AUTHOR]

The research area of image processing and its applications has significantly increased in the last decades. They can be found in medical imaging, military, surveillance, astronomy, sports, and many more. In this paper, we show two applications of image processing; heart rate measurement and a no-reference blur image quality metric. The heart rate measurement is obtained from video sequences. However, the heart rate measurement consists of artifacts and noises. Therefore, a novel method using the Canonical Component Analysis (CCA) is introduced to eliminate the artifacts and noises using two five-second video frames where one video is delayed by one second. In a no-reference blur image quality metric, we show that some exact Zernike moments (EZMs) closely exhibit the human visual system in assessing the quality of images distorted by various degrees of Gaussian blur. These EZMs are then combined through a formulation and trained using a support vector machine to give the quality scores for images from the widely used public databases. [ABSTRACT FROM AUTHOR]

Computer vision applications under low illumination, such as night-time images, possess a difficult challenge. Night-time images suffer from the loss of visibility due to inadequate illumination, which affects the performance of outdoor computer vision systems. Thus, to improve the reliability of the outdoor computer vision systems during night-time, it is essential to enhance low illumination night-time images. In this paper, we propose a lightness enhancement method that removes the night-time color cast. The proposed method compensates for the red, green, and blue color channels of the night-time image by adding a fraction of each channel to itself. Then, it averages out the compensated color channel values with the compensated mean values. It effectively removes the night-time color cast and increases the image visibility. Experiments are conducted to evaluate the effectiveness of the proposed method, and the results are compared with two night-time image enhancement methods using objective and subjective evaluations. The results show that the proposed method produced enhanced images with better image visibility than the other two methods. [ABSTRACT FROM AUTHOR]