Ultrasonic sensing may become a useful technique for distance measurement and object detection when optical visibility is not available. However, the research on detecting multiple target objects and locating their coordinates is limited. This makes it a valuable topic. Reflection signal data obtained from a single ultrasonic sensor may be just enough for the measurements of distance and reflection strength. On the other hand, if extracted properly, a scanned set of signal data by the same sensor holds a significant amount of information about the surrounding geometries. Evaluating this dataset from a single sensor scanning can be a perfect application for convolutional neural networks (CNNs). This study proposes an imaging technique based on a scanned dataset obtained by a single low-cost ultrasonic sensor. So that images are suitable for desired outputs in a CNN, a 3D printer is converted to an ultrasonic image scanner and automated to perform as a data acquisition system for the desired datasets. A deep learning model demonstrated by this work extracts object features using convolutional neural networks (CNNs) and performs coordinate estimation using regression layers. With the proposed solution, by training a reasonable amount of obtained data, 90% accuracy was achieved in the classification and position estimation of multiple objects with the CNN algorithm as a result of converting the signals obtained from ultrasonic sensors into images.

With the development of technology, target and object recognition issues are becoming increasingly important day by day. In this article, studies on ultrasonic signal processing and object recognition techniques in the literature were examined. An integrated system design that includes ultrasonic signal data obtained from different objects has been designed to contribute to the literature. With this system design, a new method is proposed that includes signal pre-processing, feature extraction and classification processes. In addition, different machine learning and artificial neural networks algorithms are used to organize and classify the data obtained from ultrasonic signals, to obtain high-performance classification rates and to achieve comprehensive object recognition. As a result of the study carried out in this context, the data set obtained from different types of objects at different distances and angles was arranged and passed through various pre-processing and statistical feature extraction methods. In conclusion, a high-performance classification rate of 97.8% using the random forest algorithm and 99.10% using the artificial neural networks algorithm was obtained during the classification phase.

Synthetic Aperture Radar (SAR) Systems are high-tech radar systems with the ability to create long-range high resolution images for day/night and all weather conditions for discovery/surveillance purposes. The images obtained from these systems can be used for target recognition with various pattern recognition and image processing methods. In this study, intense speckle noise in the SAR images of different military vehicle types was reduced by various filtering and thresholding methods. From the pre-processed images, feature extraction was performed by using Gabor filters. In the classification phase, using the sparse representation method, the high recognition accuracy is obtained from SAR images of different vehicle types. In addition, sparse representation method is superior to Naive Bayes and KNN(k nearest neighbor), which are also used in classification stage.