Multimedia Research

Enhanced Manta-Ray Foraging Optimization Algorithm based DCNN for Lane Detection

2021-06-24T15:07:42+0530

In recent days, a developed driver support system has been introduced to enhance driving, which is considered as the renowned one. In the Advanced Driver assistive systems, Lane departure detection plays an important role and it enhances the vehicle's active safe driving. Nowadays, a decent lane detection model that is on the basis of the computer vision methods, is introduced in many studies. From a stream of videos, the lane boundaries and its radius of curvatures and lane direction is detected. This video footage is recorded from a camera mounted on the top of a vehicle. Here, the lane detection model is proposed, which concentrates more on driving assistance. In this paper, a deep learning scheme is used for a lane detection model. Here, the adopted strategies possess two main components namely lane detection and image transformation. At first, the multiple lane images are obtained by the adopted technique and it transforms the image, and it aids in classifying the training. The Deep Convolution Neural Network (DCNN) classifier is considered to detect the lane from the bird's eye view image. An Enhanced Manta-Ray Foraging Optimization Algorithm (EMRFOA) is proposed to aid the DCNN classifier with the optimal weights in this paper. Finally, the developed model is analyzed with the conventional models in that the performance of the adopted model is higher than the conventional models.

Grey Wolf Optimization and Crow Search Algorithm for Resource Allocation Scheme in Cloud Computing

2021-07-03T06:11:44+0530

The computing resources are supplied by cloud computing on basis of cloud user requirements demand. Using virtualization and distributed computing, the resource allocation model is constructed to highlight the cloud services scalability. Nevertheless, a complex problem is created by the user, to manage the demand in the on-demand resource allocation model. Hence, a novel optimization approach is developed called Grey Wolf Optimization and Crow Search Algorithm (GWO-CSA) to resolve the problem in the resource allocation model. On the basis of the availability of the resources, the tasks are executed with the aid of the virtualization concept, minimize the response time. In a distributed manner, to the virtual machine, the tasks are allocated, to balance workload in the cloud. The proposed optimization method is exploited to attain effectual resource allocation. Finally, the developed method performance showed that it attains the utmost resource consumption, utmost memory consumption, and utmost CPU utilization, least skewness.

Discrete Lion Swarm Optimization Algorithm for face recognition

2021-07-14T02:31:24+0530

In various applications, face recognition plays an important role such as identification of a person, biometrics via their Closed-Circuit Television (CCTV) cameras, identity cards and etc. Besides, various biometrics like palm print, fingerprint, iris, etc plays a significant role. Therefore, in this research, a face recognition technique is developed for the classification phase as well as feature extraction. Moreover, Discrete Lion Swarm Optimization Algorithm (DLSA) is developed Deep Belief Network (DBN) for face recognition. At first, in the database, the images experience feature extraction here the feature like m-Co-HOG, Kernel based Scale Invariant Feature Transform (K-SIFT) , besides with Active Appearance Models (AAM) features which are extracted from the image. Subsequently, developed DLSA-DBN is used for the classification. The simulation of the developed technique is performed by exploiting the CVL database. In addition, the developed model outperforms the conventional models for False Acceptance Rate (FAR), accuracy, and False Rejection Rate (FRR), correspondingly.

Archimedes Optimization Algorithm: Heart Disease Prediction

2021-06-24T15:14:16+0530

Heart diseases are the most important reasons behind the high rate of morbidity and mortality among the world's population. In clinical data analysis, heart disease prediction is represented as an important problem. Progressively, the number of data is increasing, to analyze and processing it is very difficult and specially, it turns out to be to maintain the e-healthcare data. In addition, the prediction model in machine learning is considered as a necessary feature in this paper. Thus, this work concentrates to present a novel heart disease prediction technique by means of considering particular processes such as Feature Extraction, Record, minimization of Attribute, and Classification. At first, in feature extraction, both the higher-order and statistical features are extracted. Then, minimization of attribute and record is performed; to solve the curse of dimensionality the Component analysis Principle Component Analysis (PCA) acts an important role. At last, using the Neural Network (NN) model the prediction process is performed which consumes the dimensionally minimized features. Additionally, one of the main contributions of this article is to work on accurate prediction. Therefore, for the weight optimization of NN, the meta-heuristic techniques are exploited in this work. A novel optimization algorithm named Archimedes Optimization Algorithm (AOA) is proposed which resolves the aforesaid optimization issues. At last, the outcomes of the proposed method states that its efficiency over the other conventional methods.

Hybrid Wolf Pack Algorithm and Particle Swarm Optimization Algorithm for Breast Cancer Diagnosis

2021-07-27T01:29:53+0530

In women, breast cancer is deadly disease which is increased the death rate of women. By exploiting the mammogram images, a precise and early recognition of breast cancer is a complex task. Therefore, a new breast cancer recognition technique was proposed that considered five important stages: segmentation, preprocessing, feature extraction, feature selection as well as classification. Initially, by exploiting the median filtering as well as Contrast Limited Adaptive Histogram Equalization (CLAHE), input mammogram images are preprocessed. Subsequently, through the region growing method, the preprocessed images are fed to segmentation. Then, from the segmented image, texture, geometric and gradient features are extracted. The feature vector length is higher, it is important to choose optimal features. Moreover, the optimal features chosen are performed using the proposed optimization method. After completing the selection of the optimal features, they are fed to the classification procedure including the Neural Network (NN) classifier. As an innovation, to improve the precision of diagnosis (benign as well as malignant), the NN weight is chosen optimally. The NN weight optimization and the optimal feature selection are attained using the Hybrid Wolf Pack Algorithm (WPA) and Particle Swarm Optimization (PSO) Algorithm called the Hybrid WPA-PSO algorithm. At last, the performance analysis is performed between the proposed and conventional techniques.