Ahmet Kayabasi

Computer vision-based method for classification of wheat grains using artificial neural network

BACKGROUND A simplified computer vision-based application using artificial neural network (ANN) depending on multilayer perceptron (MLP) for accurately classifying wheat grains into bread or durum is presented. The images of 100 bread and 100 durum wheat grains are taken via a high-resolution camera and subjected to pre-processing. The main visual features of four dimensions, three colors and five textures are acquired using image-processing techniques (IPTs). A total of 21 visual features are reproduced from the 12 main features to diversify the input population for training and testing the ANN model. The data sets of visual features are considered as input parameters of the ANN model. The ANN with four different input data subsets is modelled to classify the wheat grains into bread or durum. The ANN model is trained with 180 grains and its accuracy tested with 20 grains from a total of 200 wheat grains. RESULTS Seven input parameters that are most effective on the classifying results are determined using the correlation-based CfsSubsetEval algorithm to simplify the ANN model. The results of the ANN model are compared in terms of accuracy rate. The best result is achieved with a mean absolute error (MAE) of 9.8 × 10-6 by the simplified ANN model. CONCLUSION This shows that the proposed classifier based on computer vision can be successfully exploited to automatically classify a variety of grains.

Millimetre wave isar imaging technique based on sparse aperture data collection

The millimetre wave (MW) applications has become very popular in recent years due to the high-resolution requirement in inverse synthetic aperture radar (ISAR) imaging. The most important problem encountered in MW imaging method is the high data collection requirement. Compressed sensing (CS) is often used in MW applications because it allows processing of signals with a sampling number below the Nyquist rate. However, since existing techniques used in CS take random samples from all spatial-frequency ISAR data, too many data collection probes are needed. In this study, CS based ISAR image is reconstructed by taking random samples from only synthetic aperture data instead of all spatial-frequency ISAR data. So, this type of data collection mechanism offers a much more practical application area for CS based ISAR imaging. The proposed method was verified by simulation results and the quality of the images were evaluated calculating ISLR.

A generalized formula in calculation of the resonant frequency of notch antenna

Notch antenna is constructed by slotting an edge of rectangular patch placed on a substrate over ground plane. Analysis of the notch antenna is complicated and very difficult due to having non-uniform shape. In this work, a novel formulation is proposed for calculating the resonant frequency of the notch antennas. The formulation regarding the resonant length of the antenna reflecting the impact of the slot is derived using Particle Swarm Optimization (PSO). Data vector of 96 notch antennas consisting of seven geometrical and electrical parameters is acquired by simulations. A resonant length formula enclosing those parameters accompanying with optimization variables is constituted in conformity with simulation data. The variables are then optimally determined by fitting the calculated resonant frequency to the simulated one by PSO algorithm. The proposed formulation is verified with simulated/measured data and validated with a test notch antenna fabricated in this study. The results demonstrate that the resonant frequency of the notch antenna can be simply calculated using the proposed formulation without dealing with sophisticated mathematics and performing simulations or measurement.