Algorithm for military object detection using image data

— The detection became difficult when the temperature difference between the tested objects and their background was very small — It was also difficult to determine the exact object´s shape due to its uneven temperature distribution. Better results were achieved with the method using a variable threshold value, depending on the given size of the object — An object detection in the visible spectrum is more difficult than object detection in the IR spectrum due to military camouflage techniques — The other method based on edge detection showed better results for detection at a long distance — Two techniques, frame difference and approximate median filter, were described for video sequence processing. The frame difference method has two main advantages + filter out disturbances in the background (waving tree branches or slow movement of uninteresting objects) + it can adapt faster to changes in the background than the other method. — Using dilation technique for objects detection in video sequences improved the detection results and could also reduce the errors caused by detection of several parts of a single object — The template matching technique could be implemented when using an R component or a grayscale image.