Image-to-image translation (IIT) has made much progress recently with the development of adversarial learning. In most of the recent work, an adversarial loss is utilized to match the distributions of the translated and target image sets. However, this may create artifacts if two domains have different marginal distributions, for example, in uniform areas. In this work, we propose an unsupervised IIT method that preserves the uniform regions after the translation. The gradient adjustment loss, which is the L2 norm between the Sobel response of the target image and the adjusted Sobel response of the source images, is utilized. The proposed method is validated on the jellyfish-to-Haeckel dataset, which is prepared to demonstrate the mentioned problem, which contains images with different background distributions. We demonstrate that our method obtained a performance gain compared to the baseline method qualitatively and quantitatively, showing the effectiveness of the proposed method.

Facial expressions of laboratory mice provide important information for pain assessment to explore the effect of drugs being developed for medical purposes. For automatic pain assessment, a mouse face tracker is needed to extract the face regions in videos recorded in pain experiments. However, since the body and face of mice are the same colour and mice move fast, tracking their face is a challenging task. In recent years, with their ability to learn from data, deep learning provides effective solutions for a wide variety of problems. In particular, convolutional neural networks (CNNs) are very successful in computer vision tasks. In this study, a CNN based tracker network called MFTN is proposed for mouse face tracking. CNNs are good at extracting hierarchical features from the training dataset. High‐level features contain semantic features and low‐level features have high spatial resolution. In the proposed MFTN architecture, target information is extracted from a combination of low‐ and high‐level features by a sub‐network, namely the Feature Adaptation Network (FAN), to achieve a robust and accurate tracker. Among the MFTN versions, the MFTN/c tracker achieved an accuracy of 0.8, robustness of 0.67, and a throughput of 213 fps on a workstation with GPU.

In this letter, a two-stage method for airport detection on remote sensing images is proposed. In the first stage, a new algorithm composed of several line-based processing steps is used for extraction of candidate airport regions. In the second stage, the scale-invariant feature transformation and Fisher vector coding are used for efficient representation of the airport and nonairport regions and support vector machines employed for classification. In order to evaluate the performance of the proposed method, extensive experiments are conducted on airports around the world with different layouts. The measures used in the evaluation are accuracy, sensitivity, and specificity. The proposed method achieved an accuracy of 94.6%, which was benchmarked with two previous methods to prove its superiority.

This letter introduces a new method for building extraction in satellite images. The algorithm first identifies the shadow segments on an oversegmented image, and then neighboring shadow segments, which are assumed to be cast by a single building, are merged. Next, candidate regions where buildings most likely occur are detected by using these shadow regions. Along with this information, closeness to shadows in illumination direction and spectral properties of segments are used to classify them as belonging to a "building" or not. Then, a resegmentation is performed by merging only the neighboring segments, which are classified as building. Finally, postprocessing is performed to eliminate some false building segments. The approach was tested on several Google Earth images, and the results are found to be promising.

The automatic detection of airports is essential due to the strategic importance of these targets. In this letter, a runway detection method based on textural properties is proposed since they are the most descriptive element of an airport. Since the best discriminative features for airport runways cannot be trivially predicted, the Adaboost algorithm is employed as a feature selector over a large set of features. Moreover, the selected features with corresponding weights can provide information on the hidden characteristics of runways. Thus, the Adaboost-based selected feature subset can be used for both detecting runways and identifying their textural characteristics. Thus, a coarse representation of possible runway locations is obtained. The performance of the proposed approach was validated by experiments carried on a data set of large images consisting of heavily negative samples.

The aim of this study is to automatically differentiate railroads from roads obtained by a general road extraction algorithm. In high resolution aerial images, the sleepers between railroad tracks depict a distinctive texture from other elongated objects. By employing this additional cue, a new Fourier based feature descriptor is developed to distinguish railroad objects in aerial images. After local Fourier features calculation, the periodic railroad texture is locally detected in the frequency domain.

In this paper, we present a novel automatic approach based on local shape descriptors to discriminate 3-D facial scans of different individuals. Our approach begins with registration, smoothing and uniform resampling of 3-D face data. Then, uniformly resampled 3-D face data are used to generate shape index, curvedness, gaussian and mean curvature values on each point of the data. Hence we obtain 2-D matrices of shape index, curvedness, gaussian and mean curvature values representing 3-D geometry information. SIFT descriptors are applied to 2-D matrices and high dimensional feature vector having shape information is obtained. Finally, high dimensional feature vector is projected to the low dimensional subspace where projection matrix is calculated by linear discriminant analysis. Features in this low dimensional subspace are compared by using cosine distance similarity metric. Proposed method is shown to have 98.35% and 98.25% detection rates at 0.001 false alarm rate for All vs. All and ROC3 experiments respectively on FRGC v2.0 database. To the best of our knowledge, these are the best results among similar studies available in 3-D face recognition literature.

A digital integrated circuit (IC) is realized using the random neural network (RNN) model introduced by Gelenbe. The RNN IC employs both configurable routing and random signaling. In this paper we present the networking/routing aspects as well as the performance results of an RNN network implemented by the RNN IC. In the RNN model, each neuron accumulates arriving signals and can fire if its potential at a given instant of time is strictly positive. Firing occurs at random, the intervals between successive firing instants following an exponential distribution of constant rate. When a neuron fires, it routes the generated pulses to the output lines in accordance with the connection probabilities. The number of neurons in the network is programmable and could be connected to each other with any desired neuron interconnection and this connection could be changed on the fly. The RNN chip architecture is cascadable to generate any network topology. All the parts of the RNN circuit are implemented using a standard digital Complimentary-Metal-Oxide-Semiconductor (CMOS) process.

In this study a new algorithm for 3D surface reconstruction from multiple images using a modified photometric stereo method is proposed and tested. The new algorithm, Filtered Lambert Photometric Stereo (FLPS), determines the non-Lambert pixels in the available images using a linearity test and constructs filtering masks for each image that corresponds to specular and self or cast shadow regions. Then, the photometric stereo is applied after eliminating the points in these masks. Tests carried out on synthetic images show that LPS on filtered images is a feasible solution when more than 4 images are available.

In the control of distillation columns, on-line composition measurements offer challenges. In this study, in order to predict the product compositions in an industrial multi-component distillation column from available on-line temperature measurements, two state estimators, an artificial neural network (ANN) and an adaptive neuro-fuzzy inference system (ANFIS), are developed and tested by using an unsteady-state column simulator. A model predictive controller (MPC) is used with the developed estimators individually for the dual composition control of the column. The performances of the developed inferential control system utilizing the estimators are found to be satisfactory considering both set-point tracking and disturbance rejection cases.

Fingerprint classification is an important part of an identification system and there are various approaches for solving this problem. In any pattern classification system the most important part is the feature extraction process. For fingerprints the orientation map is the most common and promising feature to be used for classification. Two different orientation map extraction algorithms are compared to find the one that gives the most reliable features for the PCA-MSOM classifier.

In this paper, a neural network structure based on self organizing feature maps (SOFM) is proposed for fingerprint classification. In order to be able to deal with fingerprint images having distorted regions, the SOFM learning and classification algorithms are modified. For this purpose, the concept of "certainty" is introduced and used in the modified algorithms. This fingerprint classifier together with a fingerprint identifier, constitute subsystems of an automated fingerprint identification system, named HALafis. Our results show that a network that is trained with a sufficiently large and representative set of samples can be used as an indexing mechanism for a fingerprint database, so that it does not need to be retrained for each fingerprint added to the database.

A method called optimistic method with dummy locks (ODL) is suggested for concurrency control in distributed databases. It is shown that by using long-term dummy locks, the need for the information about the write sets of validated transactions is eliminated and, during the validation test, only the related sites are checked. The transactions to be aborted are immediately recognized before the validation test, reducing the costs of restarts. Usual read and write locks are used as short-term locks during the validation test. The use of short-term locks in the optimistic approach eliminates the need for the system-wide critical section and results in a distributed and parallel validation test. The performance of ODL is compared with strict two-phase locking (2PL) through simulation, and it is found out that for the low conflict cases they perform almost the same, but for the high conflicting cases, ODL performs better than strict 2PL.>

A method for concurrency control in distributed database management systems that increases the level of concurrent execution of transactions, called ordering by serialization numbers (OSN), is proposed. The OSN method works in the certifier model and uses time-interval techniques in conjunction with short-term locks to provide serializability and prevent deadlocks. The scheduler is distributed, and the standard transaction execution policy is assumed, that is, the read and write operations are issued continuously during transaction execution. However, the write operations are copied into the database only when the transaction commits. The amount of concurrency provided by the OSN method is demonstrated by log classification. It is shown that the OSN method provides more concurrency than basic timestamp ordering and two-phase locking methods and handles successfully some logs which cannot be handled by any of the past methods. The complexity analysis of the algorithm indicates that the method works in a reasonable amount of time.>