Đang chuẩn bị liên kết để tải về tài liệu:
Swarm Robotics, From Biology to Robotics Part 11

Tham khảo tài liệu 'swarm robotics, from biology to robotics part 11', kỹ thuật - công nghệ, cơ khí - chế tạo máy phục vụ nhu cầu học tập, nghiên cứu và làm việc hiệu quả | Visual Analysis of Robot and Animal Colonies 93 binary images each color channel generates a gray-level image which is segmented by the selected method obtaining a different binary image or from a gray-level image obtained directly from the color one. Thus the first step is to convert the captured color image to a gray-level one. For such preprocessing the Hue-Saturation-Intensity HSI system is used since it encodes color information by separating an overall intensity value I from two values encoding chromaticity - hue H and saturation S. HSI might also provide better support for computer vision algorithms because it is amenable to normalization for lighting and focus on the two chromaticy parameters that are more associated with the intrinsic character of a surface rather than the lightning source. Thus the resulting gray-level image is only built from the intensity value. Derivation of HSI coordinates from RGB coordinates is a common process in computer vision Shapiro and Stockman 2001 . Once the gray-scale image is available a segmentation process has to be applied on it. Although frame difference is the easiest and fastest method to detect moving objects in an image it fails when the objects are steady. This problem could be solved by taking a reference image with no objects and subtracting it from the new ones. Nevertheless a little illumination change might make the whole process fail thus an alternative algorithm should be chosen. In our case the corresponding binary image is obtained from an input gray-scale by thresholding operations as in Swistrack. This technique defines a range of brightness values in the original image pixel value greater than a threshold or lower depending on its definition belongs to the foreground and the rest of pixels are classified as background. The drawback of this method is the correct determination of the threshold. In Swistrack two different kinds of reference images are used depending on the mode in which the system is .