Automatic small target detection in infrared images of various backgrounds from various distances

Download

index.pdf

Date

2015

Author

Yardımcı, Ozan

Metadata

Show full item record

Item Usage Stats

166
views

117
downloads

Automatic detection of small targets in various backgrounds from far distances is a very challenging problem. Background clutter and small target size are the main difficulties which should be solved while reaching a very high detection performance as well as a very low computational load. In this thesis, various methods such as Top-Hat and Wavelet Transform, edge, filtering, saliency and feature based algorithms are investigated. All of the methods are compared using some realistic test scenarios, which are created synthetically. Precision, recall, processing time and number of user dependent parameters are used to evaluate the approaches. The comparative results indicate that no algorithm can detect the target with having high precision and recall at the same time in none of the scenarios. Besides, we have realized that the methods that are used in pre-processing, detection, thresholding and post-processing stages of the algorithms are very effective on the final results. Thus, the methods used in these stages are evaluated separately and the best approach for each stage is verified. Finally, an algorithm is constructed which constitutes of the best approach for each stage. However, although we end up with a very high precision rate such as 100%, the recall values are low. In this context, a postprocessing method is proposed which increases the recall value while keeping the precision at 100% in prepared test scenarios. It is analyzed and indicated that the proposed post-processing method increases the recall value averagely 130% in all prepared test scenarios.

Subject Keywords

Infrared imaging., Imaging systems., Infrared technology., Automatic target detection.

URI

http://etd.lib.metu.edu.tr/upload/12618971/index.pdf
https://hdl.handle.net/11511/24887

Collections

Graduate School of Natural and Applied Sciences, Thesis

Suggestions

OpenMETU
Core

Automatic small target detection in synthetic infrared images YARDIMCI, Ozan; Ulusoy, İlkay (2017-04-11) Automatic detection of targets from far distances is a very challenging problem. Background clutter and small target size are the main difficulties which should be solved while reaching a high detection performance as well as a low computational load. The pre-processing, detection and post-processing approaches are very effective on the final results. In this study, first of all, various methods in the literature were evaluated separately for each of these stages using the simulated test scenarios. Then, a ...
Dim point target tracking in infrared image sequences with low SNR Güler, Zahide Selin; Vural, Elif; Department of Electrical and Electronics Engineering (2016) Dim point target tracking in infrared (IR) images has been an active research area especially in military fields. Due to the long distance from IR sensors, target appears as a dim moving point hidden in a heavily cluttered background and causes the Signal to Noise Ratio (SNR) to be very low. We present a tracking algorithm based on Particle Filters (PF), which estimates the target position by using both brightness level and motion model measurements. A target candidate list for the presented PF algorithm is...
Robust Automatic Target Recognition in FLIR imagery Soyman, Yusuf (2012-04-24) In this paper, a robust automatic target recognition algorithm in FLIR imagery is proposed. Target is first segmented out from the background using wavelet transform. Segmentation process is accomplished by parametric Gabor wavelet transformation. Invariant features that belong to the target, which is segmented out from the background, are then extracted via moments. Higher-order moments, while providing better quality for identifying the image, are more sensitive to noise. A trade-off study is then perform...
Expectation Maximization-Based Detection in Range-Heterogeneous Weibull Clutter Doyuran, Ulku Cilek; Tanık, Yalçın (2014-10-01) The problem of radar target detection in Weibull and range-heterogeneous clutter is considered. The clutter component in each range cell is modeled as an instance of a random variable belonging to one of several Weibull distributions. We use the expectation-maximization algorithm to estimate parameters of the distribution and set the threshold accordingly. The performance of the proposed receiver is investigated through computer simulations and observed to be superior compared with numerous well-known methods.
Moving hot object detection in airborne thermal videos Kaba, Utku; Akar, Gözde; Department of Electrical and Electronics Engineering (2012) In this thesis, we present an algorithm for vision based detection of moving objects observed by IR sensors on a moving platform. In addition we analyze the performance of different approaches in each step of the algorithm. The proposed algorithm is composed of preprocessing, feature detection, feature matching, homography estimation and difference image analysis steps. First, a global motion estimation based on planar homography model is performed in order to compensate the motion of the sensor and moving ...

Citation Formats

O. Yardımcı, “Automatic small target detection in infrared images of various backgrounds from various distances,” M.S. - Master of Science, Middle East Technical University, 2015.