Vehicle detection and classification are important tasks in the automatic traffic monitoring system. The proposed work focuses on vehicle detection and classification. Vehicle detection is carried out using the combination of dense optical flow method and integrated binary projection profile. Inception-ResNet-v2 is used as a feature extraction technique and extracted features are fed to two different classifiers such as Support Vector Machine and Random Forest to classify the vehicle type. The recognition performance of Inception-ResNet-v2 with these classifiers is significantly high and the proposed approach obtained an output accuracy as 99.89% and 98.615% in Support Vector Machine and Random forest respectively.

This paper proposes a method for estimating the joint direction of departure (DOD) and direction of arrival (DOA) of coherent targets in multiple-input multiple-output (MIMO) radar with electromagnetic vector sensors. The method does not require any additional processing for both decorrelating the coherent targets and pairing the DOD and DOA, and exhibits real-valued operations. Compared to the existing polarization smoothing and unitary ESPRIT, the proposed method shows improved performance which is validated by the simulation results.

This study proposes a novel integrated polarisation and diversity smoothing (IPDS) algorithm to solve the direction estimation problem of coherent targets in bistatic multiple‐input multiple‐output (MIMO) radar with electromagnetic vector sensors (EVSs). The IPDS algorithm partitions the transmit and receive arrays into multiple subarrays based on the sensor elements and the spatial phase shift factors. Then the covariance matrices of the received signals from coherent targets associated with these subarrays are smoothened to restore the rank inadequacy of the reflection coefficient covariance matrix. After applying the IPDS algorithm, the direction of departure (DOD) and the direction of arrival (DOA) are estimated using the estimation of signal parameters via rotational invariance technique and the multiple signal classification methods. The effectiveness of the proposed algorithm in terms of decorrelation factor, estimation accuracy, precision and the spatial spectrum is evaluated through computer simulations. Cramér–Rao lower bound is derived for the bistatic MIMO radar with EVSs to substantiate the proposed algorithm. The simulation results are compared with the diversity smoothing and the polarisation smoothing algorithms available in literature. The results obtained by using the proposed algorithm were found to be impressive.

In this paper, a novel technique for the removal of both horizontal and vertical lines in the document images are presented using shift, difference and coordinate logic operations. The method is also employed in the non-document images to remove the lines and to detect the edges. The proposed method shows promising results than the conventional methods using morphological operations for edge detection. The performance of the proposed method is demonstrated by considering printed, handwritten document and non-document gray level as well as coloured images.