Large-scale psychophysical experiments are carried out on two types of mobile displays to evaluate the perceived image quality (IQ). Eight perceptual attributes, i.e., naturalness, colorfulness, brightness, contrast, sharpness, clearness, preference, and overall IQ, are visually assessed via categorical judgment method for various application types of test images, which were manipulated by different methods. Their correlations are deeply discussed, and further factor analysis revealed the two essential components to describe the overall IQ, i.e., the component of image detail aspect and the component of color information aspect. Clearness and naturalness are regarded as two principal factors for natural scene images, whereas clearness and colorfulness were selected as key attributes affecting the overall IQ for other application types of images. Accordingly, based on these selected attributes, two kinds of empirical models are built to predict the overall IQ of mobile displays for different application types of images.