Wound assessment using a smartphone image has recently emerged as a novel way to provide actionable feedback to patients and caregivers. Wound segmentation is an important step in image-based wound assessment, after which the wound area can be analyzed. Semantic segmentation algorithms for wounds assume favorable lighting conditions. However, smartphone wound imaging in natural environments can encounter adverse lighting that can cause several errors during semantic segmentation of wound images, which in turn affects the wound analysis. In this work, we study and characterize the effects of adverse lighting on the accuracy of semantic segmentation of wound images. Our findings inform a deep learning-based approach to mitigate the adverse effects. We make three main contributions in this work. First, we create the first large-scale Illumination Varying Dataset (IVDS) of 55440 images of a wound moulage captured under systematically varying illumination conditions and with different camera types and settings. Second, we characterize the effects of changing light intensity on U-Net’s wound semantic segmentation accuracy and show the luminance of images to be highly correlated with the wound segmentation performance. Especially, we show low-light conditions to deteriorate segmentation performance highly. Third, we improve the wound Dice scores of U-Net for low-light images to up to four times the baseline values using a deep learning mitigation method based on the Retinex theory. Our method works well in typical illumination levels observed in homes/clinics as well for a wide gamut of lighting like very dark conditions (20 Lux), medium-intensity lighting (750 - 1500 Lux), and even very bright lighting (6000 Lux).

Forty-two college aged participants took part in a mixed repeated measures factorial design experiment that assessed color memory as a function of condition (practice with feedback, practice without feedback and no practice), dimension (hue, saturation and lightness) and color (red, yellow, green and blue). Attention was focused on the distinction between memory color and color memory, color experience and preference, mechanisms of color perception and theories of color vision (see below). Only two significant effects were found: a significant main effect for dimension and a significant interaction between dimension and color. Pearson correlations were assessed between color memory and color experience, color preference and observer imagery. None of the correlations were significant. The results of the experiments revealed that practice does not have a significant effect on color memory and the conclusion, therefore, is that the phenomenon of color memory is not improved by practice. A tentative explanation involves the early stages of color processing which are presumed to be computational in nature and to take place independently of cognitive processes such as learning and memory, which do not take place until visual information has reached the extrastriate areas. By that time, color information has been combined with information about context, in area V4 of the human visual cortex (Zeki & Marini, 1998). Although it has been shown through this experiment that practice does not improve memory for color, the possibility remains that practice may improve memory color for specific objects - namely ecologically relevant stimuli - since memory color involves higher order processing, such as learning and memory. / Master of Science

Many applications require automatic image analysis for different quality of the input images. In many cases, the quality of acquired images is suitable for the purpose of the application. However, in some cases the quality of the acquired image has to be modified according to needs of a specific application. A higher quality of the image can be achieved by Image Enhancement (IE) algorithms. The choice of IE technique is challenging as this choice varies with the application purpose. The goal of this research is to investigate the possibility of the selective application for the IE algorithms. The values of entropy and Peak Signal to Noise Ratio (PSNR) of the acquired image are considered as parameters for selectivity. Three algorithms such as Retinex, Bilateral filter and Bilateral tone adjustment have been chosen as IE techniques for evaluation in this work. Entropy and PSNR are used for the performance evaluation of selected IE algorithms. In this study, we considered the images from three fingerprint image databases as input images to investigate the algorithms. The decision to enhance an image in these databases by the considered algorithms is based on the empirically evaluated entropy and PSNR thresholds. Automatic Fingerprint Identification System (AFIS) has been selected as the application of interest. The evaluation results show that the performance of the investigated IE algorithms affects significantly the performance of AFIS. The second conclusion is that entropy and PSNR might be considered as indicators for required IE of the input image for AFIS.

We argue that one of the early goals of color vision is to distinguish one kind of material from another. Accordingly, we show that when a pair of image regions is such that one region has greater intensity at one wavelength than at another wavelength, and the second region has the opposite property, then the two regions are likely to have arisen from distinct materials in the scene. We call this material change circumstance the 'opposite slope sign condition.' With this criterion as a foundation, we construct a representation of spectral information that facilitates the recognition of material changes. Our theory has implications for both psychology and neurophysiology. In particular, Hering's notion of opponent colors and psychologically unique primaries, and Land's results in two-color projection can be interpreted as different aspects of the visual system's goal of categorizing materials. Also, the theory provides two basic interpretations of the function of double-opponent color cells described by neurophysiologists.