Size uniformity is one of the prominent features of superpixels. However, size uniformity rarely conforms to the varying content of an image. The chosen size of the superpixels therefore represents a compromise - how to obtain the fewest superpixels without losing too much important detail. We present an image segmentation technique that generates compact clusters of pixels grown sequentially, which automatically adapt to the local texture and scale of an image. Our algorithm liberates the user from the need to choose of the right superpixel size or number. The algorithm is simple and requires just one input parameter. In addition, it is computationally very efficient, approaching real-time performance, and is easily extensible to three-dimensional image stacks and video volumes. We demonstrate that our superpixels superior to the respective state-of-the-art algorithms on quantitative benchmarks.

We propose a daltonization method that enhances chromatic edges and contrast for color-deficient people by optimizing the gradient of an image. We rotate and scale the error gradient between the original and its simulation in the color space into the direction of optimal visibility that is orthogonal to both the main direction of information loss and the direction of lightness. Then, we reintegrate the daltonized image version from the modified gradient through an iterative diffusion process. Moreover, we include multiscaling to guarantee optimal daltonization on different scales of the image. Also, we provide an interface for data attachment modules designed to maintain naturalness of memory colors like neutral colors. We evaluate and compare our proposed method to other top-performing daltonization methods in behavioral and psychometric experiments. A visual-search experiment assessing performance of the attentional mechanism of the human visual system before and after daltonization measures the greatest improvement in accuracy for our proposed method compared to the original and all investigated daltonization methods. It also reveals optimal results for both natural and Ishihara images among both protan and deutan color-deficient observers. Furthermore, we can deduce from the results of a pairwise preference evaluation that our proposed method is preferred highest amongst all color-deficient people in total. Our proposed method is also ranked among the most preferred daltonization methods for both protan and deutan color-deficient observers individually.

Texture features can be considered as methods for encoding an image: taking pixel intensities or filter responses and forming them into a description which can be used to solve problems including recognition and matching. In this paper we are considering the inverse problem: given a textural representation of an image, how well can we recover the original. We show how the LBP method encodes comparative relationships between pixels and relative to these relations we can recover an image with our new method. We extend the recovery method to work with the Sudoku texture representation (an extension of LBP). We show that this method produces a reconstruction more correlated with the original image than the prior art.

OLED display technology is gaining popularity among original equipment manufacturers (OEM). Production costs are decreasing, making this technology more readily available. OLED displays have a better contrast, no backlight and the ability to estimate the contribution of each pixel to the power of the display. This feature allows to experiment spatial algorithms to improve the image quality in relation to its power consumption. In this article we present a framework to evaluate the performance of spatial algorithms such asjust noticeable distortion and saliency maps on OLED displays. We introduce a comprehensive power model that takes into account each pixel value and the display screen brightness. We validate the effectiveness of this model by implementing a power reduction method based on power saving and perceptual quality metric.

Scenes with back-light illumination are problematic when captured with a typical LDR camera, as they result in dark regions where details are not perceivable. In this paper, we present a method that, given an LDR backlit image, outputs an image where the information that was not visible in the dark regions is recovered without losing information in the already well-exposed parts of the image. Our method has three main steps: first, a variational model is minimized using gradient descent, and the iterates of the minimization are used to obtain a set of weight maps. Second, we consider the tone-mapping framework [3J that depends on four parameters. Two different sets of parameters are learned by dividing the image in the darker and lighter parts. Then, we interpolate the two sets of parameter values in as many sets as weighting maps, and tone-map the original image with each set of parameters. Finally, we merge the new tone-mapped images depending on the weighting maps. Results show that our method outperforms current backlit image enhancement approaches both quantitatively and qualitatively.

Color imaging involves a variety of processing operations, from interpolation, via matrix transformation, to smoothing and predictive modeling. Since colors can be represented as coordinates in color space, the general methods of mathematics can be applied to them. However, if color coordinates are treated simply as generic spatial coordinates, their processing can have undesirable consequences, deriving from a disconnect between the coordinates representing a color and the color formation properties resulting in it. E.g., interpolating among colors with very different lightnesses may lead to a grainy result in print, or varying the interpolation support when processing a transition may lead to unwanted cross-contamination of colorants. To address such challenges, the present paper proposes two color processing algorithms that do take the color properties of processed coordinates into account. They can therifore, in some sense, be thought of as "color color" processing algorithms rather than as geometric or mathematical color processing ones. The consequences of making color-native choices when processing color data then are improved transitions, "purity" and grain.

The goal of this work is to provide a guideline to produce high quality chromatic contrast sensitivity (CCS) data at low frequencies, and to contrast with the previously published data. An experiment was carried out usingforced-choice stair-case method to investigate the CCS just noticeable difference (JND) in different color changing directions at different spatial frequencies. The JND ellipses at different spatial frequencies were fitted and compared with those earlier studies. The results from a white and a green center were reported. They provided theoretical basis and standard practice for the lighting and the imaging industries.

Raw images are more useful than JPEG images for machine vision algorithms and professional photographers because raw images preserve a linear relation between pixel values and the light measured from the scene. A camera is radiometrically calibrated if there is a computational model which can predict how the raw image is mapped to the corresponding rendered image (e.g., JPEGs) and vice versa. Our method makes use of the observation that the rank order of pixel values is mostly preserved post-color correction. We show that this observation is the key for getting a compact and robust radiometric calibration model. Since our method requires fewer variables, it can be solved for using less calibration data. An additional advantage is that we can derive the camera pipeline from a single pair of raw-JPEG images. Experiments demonstrate that our method delivers state-of-the-art results (especially for the most interesting conversion from JPEG to raw). © 2018 Society for Imaging Science and Technology. © 2018 Society for Imaging Science and Technology.

Measuring the spectral responsivity of a camera using a monochromator is time-consuming and expensive. This work evaluates afast responsivity measurement method, where diffraction spectrum images are captured and then used for estimating camera responsivity. An error was noticed in the previously proposed measurement method that was caused by spectroradiometer measurement errors and vignetting effects from the camera's lens and sensor. Therefore, a correction step using chromaticity error minimisation is presented to adjust the initial responsivity estimate. It requires a chart to be captured under a known illumination. The chromaticity error of the improved procedure is approximately one order of magnitude smaller than the original error. This enhanced method was employed to create a dataset of spectral responsivities for machine vision, photographic, and movie cameras, which is presented here.