An automatic target recognation algorithm intended for hardware implementation

Automatic target recognition (ATR), is a title of general concept of the systems that designed for a particular goal and aimed to give a purposeful output by executing a specific algorithm respectively and autonomously. The machineries or autonomous structures that designed for aim of automatic target recognition are generally named as ATR systems. The ATR systems are getting to be more preferred in civilian and warfare application areas because of switching off human factors that have a narrow sense of optical spectrum and being able to discern more complex details of data.Techniques that used for designing ATR systems can be evaluated in two main topic: image based and statistical based techniques. In this work both of the techniques are searched and discussed with a wide range of literature background. However image processing techniques having steep trend on evolution with great advantages in more area are considered with application result of techniques.One of the the most crucial drawback of the autonomous systems, namely aided target recognition or automatic target recognition or automatic tracking system or ATR systems, are designing for specific areas or issues and consequently systems come with solution in a narrow specific area. Especially image processing based techniques systems are designing and using for a particular problem rather than be able to making use of a wide range of recognition areas. The most considerable reason of this downside is designing the algorithms depending on the specific problems. This issue that is a significant downside of ATR systems could be optimizing when the system is needed to use in an other area or problems make the ATR systems problem-oriented.In this study, drawbacks of the ATR systems are probed and potential solutions of the problems are suggested. Using circular sequence of direction number of Freeman chain code for rotation invariant feature of the ATR systems as a feature is one of this solutions. Another solution is suggested for using a effective edge detection techniques combining mathematical morphology edge detection and canny algorithms for more affective segmentation. Furthermore, artificial neural network are discussed in classification module on behalf of designing of a holistic atr system.Designed ATR systems are tested using both image and statistical techniques on a wide range photo database and the test result are examined. Drawbacks and advantages of the ATR systems are investigated and potential solution are presented in details.