An Artificial Intelligent Centered Object Inspection System Using Crucial Images

An Artificial Intelligent Centered Object Inspection System Using Crucial Images

Santosh Kumar Sahoo (Utkal University, Department of Electronics and Telecommunications Engineering, Bhubaneswar, India) and B. B. Choudhury (Utkal University, Department of Mechanical Engineering, Bhubaneswar, India)
Copyright: © 2018 |Pages: 14
DOI: 10.4018/IJRSDA.2018010104
OnDemand PDF Download:
No Current Special Offers


This article proposes a unique optimization algorithm like Adaptive Cuckoo Search (AdCS) algorithm followed by an Intrinsic Discriminant Analysis (IDA) to design an intelligent object classifier for inspection of defective object like bottle in a manufacturing unit. By using this methodology the response time is very faster than the other techniques. The projected scheme is authenticated using different bench mark test functions along with an effective inspection procedure for identification of bottle by using AdCS, Principal-Component-Analysis (PCA) and IDA. Due to this the projected procedures terms as PCA+IDA for dimension reduction in addition to this AdCS-IDA for classification or identification of defective bottles. The analyzed response obtained from by an application of AdCS algorithm followed by IDA and compared to other algorithm like Least-Square-Support-Vector-Machine (LSSVM), Linear Kernel Radial-Basic-Function (RBF) to the proposed model, the earlier applied scheme reveals the remarkable performance.
Article Preview


Due to rapid growth of industry to meet the society’s requirement it is necessary for maintaining a quality product. In order to achieve this, most of the manufacturing unit follow the automation scheme as a result a finished quality product can be delivered to an end user within an optimum interval. So, in most of the automation, an image processing scheme is used to facilitate the model more efficient to recognize the object of interest in a smooth manner. The proposed model uses an artificial intelligent scheme for recognition of defective bottle in a manufacturing unit where the different algorithm and techniques are used to validate the suggested plan. Before object identification it is highly necessary to reduce the dimension of the captured image by an application of different linear tools such as PCA, LDA as well as some nonlinear tools like Artificial neural network scheme (ANN), isometric mapping, locally linear embedding and Laplacian Eigen maps etc. But apart from this the linear tools are the best choice because the nonlinear tools required heavy computational work for different parameter tuning and lack of handling capacity for testing data as compared to linear one. The PCA estimates the highest width data into acouple of base function to catch a squeezed demo of the initial data as a result dimension decline can be realized and it mainly reflects the Eigen faces of the detected image features. Likewise, LDA reflects the features as a Fisher faces in which unlike class’s data points are mapped with an optimum gap or distance. The IDA scheme is utilized for maximizing uniqueness dissimilarities and minimizing intra personal alteration. Again the selection procedure for optimal Eigen vector is inaccurate. So, to overcome this problematic condition the PCA + IDA algorithm remains used for increasing the performance level then this performance is compared with the other identification techniques like KTHNN, ANN, SVM and LSSVM method. Finally, it is concluded that the projected scheme be able to classify impaired bottle promptly as well as accurately. Therefore, the curative exploit determination be continuing on a crucial stage of manufacturing progression so that the ruinquality risk be adjusted.

Complete Article List

Search this Journal:
Volume 8: 1 Issue (2022): Forthcoming, Available for Pre-Order
Volume 7: 4 Issues (2021): 1 Released, 3 Forthcoming
Volume 6: 3 Issues (2019)
Volume 5: 4 Issues (2018)
Volume 4: 4 Issues (2017)
Volume 3: 4 Issues (2016)
Volume 2: 2 Issues (2015)
Volume 1: 2 Issues (2014)
View Complete Journal Contents Listing