This paper proposes a salient chromagram components extraction method based on the temporal discrete cosine transform of a chromagram block to improve cover song retrieval accuracy. The proposed salient chromagram emphasizes tonal contents of music, which are well-preserved between an original song and its cover version, while reducing the effects of timbre difference. We apply the proposed salient chromagram extraction method as a preprocessing step for the Fourier-transform based cover song matching. Experiments on two cover song datasets confirm that the proposed salient chromagram improves the cover song search accuracy.

Image denoising is considered as a key factor for capturing high-quality photos in digital cameras. Thus far, several image denoising methods have been proposed in the past decade. In addition, previous studies either relied on deep learning-based approaches or used the hand-crafted filters. Unfortunately, the previous method mostly emphasized on image denoising regardless of preserving or recovering the detail information in result images. This study proposes an detail extraction network to estimate detail information from a noisy input image. Moreover, the extracted detail information is utilized to enhance the final denoised image. Experimental results demonstrate that the proposed method can outperform the existing works by a subjective measurement.

In this paper, we propose a novel way of combining multiple deep convolutional neural network (DCNN) architectures which work well for accurate video face identification by adopting a serial combination of 3D and 2D DCNNs. The proposed method first divides an input video sequence (to be recognized) into a number of sub-video sequences. The resulting sub-video sequences are used as input to the 3D DCNN so as to obtain the class-confidence scores for a given input video sequence by considering both temporal and spatial face feature characteristics of input video sequence. The class-confidence scores obtained from corresponding sub-video sequences is combined by forming our proposed class-confidence matrix. The resulting class-confidence matrix is then used as an input for learning 2D DCNN learning which is serially linked to 3D DCNN. Finally, fine-tuned, serially combined DCNN framework is applied for recognizing the identity present in a given test video sequence. To verify the effectiveness of our proposed method, extensive and comparative experiments have been conducted to evaluate our method on COX face databases with their standard face identification protocols. Experimental results showed that our method can achieve better or comparable identification rate compared to other state-of-the-art video FR methods.

In this paper, we compare and analyze the classification performance of deep learning algorithm Convolutional Neural Network(CNN) ac cording to ensemble generation and combining techniques. We used several CNN models(VGG16, VGG19, DenseNet121, DenseNet169, DenseNet201, ResNet18, ResNet34, ResNet50, ResNet101, ResNet152, GoogLeNet) to create 10 ensemble generation combinations and applied 6 combine techniques(average, weighted average, maximum, minimum, median, product) to the optimal combination. Experimental results, DenseNet169-VGG16-GoogLeNet combination in ensemble generation, and the product rule in ensemble combination showed the best performance. Based on this, it was concluded that ensemble in different models of high benchmarking scores is another way to get good results.

Wireless Sensor Networks (WSNs) consist of multiple tiny and power constrained sensors which use radio frequencies to carry out sensing in a designated sensor area. To effectively design and implement reliable WSN, it is critical to consider models, protocols, and algorithms that can optimize energy consumption of all the sensor nodes with optimal amount of packet delivery. It has been observed that deploying a single sink node comes with numerous challenges especially in a situation with high node density and congestion. Sensor nodes close to a single sink node receive more transmission traffic load compared to other sensors, thus causing quick depletion of energy which finally leads to an energy hole and sink hole problems. In this paper, we proposed the use of multiple energy efficient sink nodes with brute force technique under optimized parameters to improve on the number of packets delivered within a given time. Simulation results not only depict that, deploying N sink nodes in a sensor area has a maximum limit to offer a justifiable improvement in terms of packet delivery ratio but also offers a reduction in End to End delay and reliability in case of failure of a single sink node, and an improvement in the network lifetime rather than deploying a single static sink node.

Information spread in networks is universal in many real-world phenomena such as propagation of infectious diseases, diffusion of a new technology, computer virus/spam infection in the internet, and tweeting and retweeting of popular topics. The problem of finding the information source is to pick out the true source if information spread. It is of practical importance because harmful diffusion can be mitigated or even blocked e.g., by vaccinating human or installing security updates. This problem has been much studied, where it has been shown that the detection probability cannot be beyond 31% even for regular trees if the number of infected nodes is sufficiently large. In this paper, we study the impact of an anti-information spreading on the original information source detection. We consider an active defender in the network who spreads the anti-information against to the original information simultaneously and propose an inverse Voronoi partition based inference approach, called Voronoi Inference to find the source. We perform various simulations for the proposed method and obtain the detection probability that outperforms to the existing prior work.

The level of psychological damage caused by verbal abuse among cyberbully victims is very serious. It is going to introduce a system that determines the level of sanctions against chatting in real time using the automatic prohibited words filtering based on artificial neural network. In this paper, we propose a keyword filtering method that detects the modified prohibited words and determines whether the corresponding chat should be sanctioned in real time, and a real-time chatting screening system using it. The accuracy of filtering through machine learning was improved by processing data in advance through coding techniques that express consonants and vowels of similar pronunciation at close distances. After comparing and analyzing Mahalanobis-based clustering algorithms and artificial neural network-based algorithms, algorithms that utilize artificial neural networks showed high performance. If it is applied to Internet chatting, comments or online games, it is expected that it will be able to filter more effectively than the existing filtering method and that this will ease communication inconvenience due to existing indiscriminate filtering methods.

This research implemented grammatical rules for relationship extraction from class diagram candidate. The problem statement is generated by our algorithm to yield class diagram and candidate relationship candidates. The relationships of class diagrams are extracted automatically from the problem statement by using Natural Language Processing (NLP). The extraction used the grammatical rules that obtained from various sources and translated into our algorithm. The performance evaluation of the extraction algorithm used ATM problem statements. The application captures the problem statement and draws automatically the relations of class diagrams using Forced-Directionality Graph algorithm. The performance evaluations show refining methods for class diagram and relationships extraction improve recall score.

This study explores the concept of spectacular realism. and finds out how spectacular realism is made in movies. The Movies's spectacle is a realistic spectacle that does not miss the reality. so audiences can immerse themselves in it. Nowadays the spectrum of visual realism is becoming very diverse. "Perceptual realism" means to feel more like real than to reproduce reality as it is. "Synthetic realism" is a realism that CGI makes. And "Spectacular realism" makes the spectators concentrate to images than narrative or character. It means the dominance of the visual. The films and showed "Spectacular realism" especially in the aerial battle scenes. The films use 'Presentation of a new spatial sensation using unusual angles', 'Editing that constitutes transcendental time' and 'Intense contrast' as methods to embody spectacular realism. And the harmonious arrangement of transcendental images and very realistic screens effectively embodies Spectacular Realism. This discussion, which began in the air battle scene, could contribute to a broader and more diverse exploration of the implementation of spectacular realism images in the future.

This study defined the animation space as the concept of map based on the space of Joseph Campbell's heroic narrative. The space in which the animated character exists is an image symbolic language that reflects the inner and outer growth stages of the hero. I focused on the spatiality that plays the role of the power to lead the narrative and examined the meaning as the mediator that leads the heroic narrative. First, 6 spatial variables were derived by observing 'Hero's Journey 'which is used as a basis of US commercial animation scenario. Second, spatial variables are defined as 64 zones and proposed as 'Spatial Map Model of the Hero's Journey' (SMMH). Third, the character space of was applied to 'SMMH', and the change of space utilization rate and spatiality and the narrative meaning were analyzed. This study extended the narrative of animation space which was not actively studied to map concept. It is possible to provide a different viewpoint in animation production and research.

Both within the country and internationally, discussions about core technologies that will be prominent in future society have been initiated, and software education for the rapidly changing industrial framework has been essential and implemented in earnest. Also, there is an increasing interest in interactive technology and software education where people and people, people and computers, and computers and objects communicate with each other in today's day-to-day life, connected through the digital world. Therefore, it is necessary to investigate the user's field of interest and the best technical education method for teaching how to operate the production tool for implementing the interactive technology. In this study, we utilize the Processing program which is widely used as interactive technology software training and creativity tool to improve the understanding of the interactive technology related to the 4th industrial revolution. This study aims to determine and propose the best method of using a Processing library that can be applied to each major field.