Coded-Aperture Computational Millimeter-Wave Image Classifier Using Convolutional Neural Network

Abstract

A millimeter-wave (mmW) classifier system applied to images synthesized from a codedaperture based computational imaging (CI) radar is presented. A developed physical model of a CI system is used to generate the image dataset for the classification algorithm. A convolutional neural network (CNN) is integrated with the physical model and trained using the dataset comprising of synthesized mmW images obtained directly from the developed CI physical model. A k-fold cross validation technique is applied during the training process to validate the classification model. The coded-aperture CI concept enables image reconstruction from a significantly reduced number of back-scattered measurements by facilitating physical layer compression. This physical layer compression can substantially simplify the data acquisition layer of imaging radars, which is realized using only two channels in this article. The integration of the classification algorithm with the CI numerical model is particularly important in enabling the training step to be carried out using relevant system metrics and without the necessity for experimental data. Leveraging the CI numerical model generated data, training step for the classification algorithm is achieved in real-time while also confirming that the numerically trained CI classifier offers high accuracy with both simulated and experimental data. The classifier integrated physical model also enables performance analysis of the classification algorithm to be carried out as a function of key system metrics such as signal-to-noise (SNR) level, ensuring a complete understanding of the classification accuracy under different operating conditions. The trained CI system is tested with synthesized mmW images from the physical model and a classification accuracy of 89% is achieved. The proposed model is also verified using experimental data validating the fidelity of the developed CI integrated classifier system. A classification latency of 3.8 ms per frame is achieved, paving the way for real-time automated threat detection (ATD) for security-screening applications.INDEX TERMS Millimetre-wave, imaging radars, computational imaging, neural networks, image classification, coded-aperture.

Intelligent meta-imagers: From compressed to learned sensing

“…More recently, various machine learning and deep learning algorithms have been applied in computational imaging [126][127][128][129][130] and inverse scattering, 131,132 including some cases of compressive metaimagers. 58,82,83,97,[133][134][135] Their potential benefits include fast online inference on tasks that are difficult or impossible to formulate analytically, at the cost of expensive offline training accompanied by a need for a large training dataset. With reference to the sensing pipeline from Fig.…”

3-D Super-Resolution of Coded Aperture Millimeter-Wave Images Using Complex-Valued Convolutional Neural Network

“…Second, because the training size is increased, the computation time and memory requirement for the training process will also increase. To overcome these issues, we have developed a physical model to simulate a mmW radar imaging framework [11], [12]. The imaging model is based on the computational imaging (CI) coded aperture technique [13]- [16].…”

“…Fig. 3 shows the set-up of the CI model [11], [12]. It consists of two dynamic apertures: transmitter and receiver, operating in a bi-static mode.…”

“…This indirect mapping of the scene information eliminates the need for raster scanning, a significant advantage of CI based imaging modalities. From [11], [12], it is known that the back-scattered measurements can be used to reconstruct back the scene information by Eq. 1.…”

“…The developed SR model in this paper can be of help in this particular step. As a part of future work, the SR model can be integrated with a classifier model that we have previously developed [11]. To this end, any input images can be passed through the SR model first before feeding it to the classifier model.…”