A salient image region is defined as an image part that is clearly different from its surround. This difference is measured in terms of a number of attributes, namely, contrast, brightness and orientation. By measuring these attributes, visual saliency algorithms aim to predict the regions in an image that would attract our attention under free viewing conditions. As the number of saliency models has increased significantly in the past two decades, one is faced with the challenge of finding a metric that can be used to objectively quantify the performance of different saliency algorithms. To address this issue in this article, first, the state of the art of saliency models is revisited. Second, the major challenges associated with the evaluation of saliency models are discussed. Third, ten frequently used evaluation metrics are examined and their results are discussed for ten latest state-of-the-art saliency models. For the analysis, a comprehensive open source fixations database has been quantitatively examined.
A control sequence gives the intensities of the primaries for a pixel in a display. A multi-primary display has four or more primaries, so that multiple control sequences can sometimes produce the same color. Different primaries likely consume different amounts of energy; furthermore, the energy consumption can be a complicated function. A minimal-energy control sequence for a target color produces that color with as little energy as possible. This article shows that such minimal-energy sequences take a simple geometric form when each primary’s energy function is linear. The display gamut, in CIE XY Z space, can be dissected into parallelepipeds. The originating vertex of each parallelepiped is the sum of a set of primaries at full intensity. Each edge of a parallelepiped is the translation of one primary. A color with XYZ coordinates in a certain parallelepiped is a unique linear combination of the primaries in the originating vertex, and the three edge primaries. This article proves that there exists a dissection such that these linear combinations are minimal-energy control sequences. In the generic case, this dissection is unique. An algorithm for a minimal-energy dissection is presented, along with an example.
Existing morphological image reconstruction methods check all the neighbors around a current pixel for the possibility of morphological reconstruction. However, the range of neighbors that can be reconstructed on the current pixel can be reduced using the directions propagated from a previously reconstructed pixel, thereby reducing the total number of checks. Therefore, this article proposes morphological image reconstruction using directional propagation that shortens the total reconstruction time by utilizing the directions propagated from a previously reconstructed pixel. On comparing the runtime for the proposed method and four state-of-the-art methods using gray images, the proposed method reduces the execution time.
Ultra-high frequency radio frequency identification tag antennas were screen printed directly onto pharmaceutical packaging and a comparison between commercially available inlays (hereafter reference tags) and printed tag antennas using two different inks (hereafter tags A and B) was made. The research revealed that the properties of the reference tag were slightly better than those for tags A and B on the empty boxes. Nevertheless, all of the samples can be successfully read for lengths of up to 105 and 125 cm for horizontally and vertically oriented tags, respectively. None of the tags compared can be read successfully when placed in close proximity to a metallic surface or a liquid.
Deterioration in reading capabilities may be prevented by inserting a material with a low dielectric constant ε. The authors proved that the readability of all of the tags is improved when additional cardboard sheets are inserted between the Al blister and the tags. The reference tags require a separation distance of at least 4 mm, whereas the tags A and B need only a separation distance of just over 2 mm to be read successfully.
The effect of liquids in glass vials is less pronounced than the effect of reflective metal. Generally, there is no need for low ε separation since the walls of the glass vial act as a separator. Regardless, the backscattered signal is approximately 5 dBm lower than the signal from the empty boxes and, in any event, the reading range for all of the tags does not fall below 60 cm.
Histogram equalization facilitates information identification in the case of a concentrated image intensity. However, this method is not suitable for consumer electronics, such as televisions, as the focus on contrast enhancement can create noise. Thus, to reduce this noise, contrast-limited adaptive histogram equalization uses a maximum threshold for contrast enhancement, but this can result in loss of image detail. Therefore, the author proposes an image enhancement method that avoids loss of image detail through the generation of detail maps based on multiple layer block overlapped histogram equalization. As a result, the loss of image detail is minimized by combining images with limited contrast enhancement using a detail map for each layer.