Effect of Enhancement on Convolutional Neural Network Based Multi-view Object Classification

Date of Award


Degree Name

M.S. in Electrical Engineering


Department of Electrical Engineering


Advisor: Vijayan Asari


The main goal of this thesis is classification of multi-view objects by using convolutional neural networks (CNN), and evaluation of the recognition performance on images preprocessed by enhancement technologies such as multilevel windowed inverse sigmoid (MWIS) function and locally tuned sine nonlinearity (LTSN) technique. Humans can easily recognize objects in different observational directions, but machines cannot achieve this easily. The convolutional neural network (CNN), which has successfully been used to do visual imagery analysis, is a deep learning, feed-forward neural network that collects features of an image and classify them accordingly. A multi-layer CNN architecture is designed for multi-view object classification by appropriately choosing the number of layers, the sequence of layers cascading, and size of the filters. It is expected that the enhanced images exhibit stronger features. Therefore, we apply image enhancement techniques before the convolutional neural network to observe the recognition performance. The datasets used for performance evaluation in this work are from the Columbia Object Image Library (COIL-100) and Multi-view Car dataset. It is observed that the preprocessing by image enhancement can provide improved performance in some cases of the smaller training set. Research work is in progress to modify the CNN architecture to see the impact of recognition performance for multi-view object classification. Advanced non-linear enhancement technologies might also be investigated to see the effectiveness in classification.


Electrical Engineering, Multi-view object classification, Convolutional neural network, Machine Learning, Multilevel windowed inverse sigmoid function, Locally tuned sine nonlinearity function

Rights Statement

Copyright 2018, author