| Summary: | 碩士 === 中華大學 === 電機工程學系碩士班 === 100 === The purpose of this paper is to achieve the recognition of guide routes by the neural network, which integrates the approaches of color space conversion, image binary, morphology, connected-component labeling, feature extraction, and feature value acquisition from Graphic User Interface Design Environment (GUI) which designed by Matlab software. Using a color image as an input, RGB color space method can’t directly provide us with the value, saturation, and hue of a color image pixel, so we employ HSV color space to convert the image. Since the image contain a lot of information, we use Otsu method based on image binary to reduce the complexity of the image and extract the information from it. Image going through binary often generates information that caused by the uneven background, thus we make use of the idea of dilation and erosion from morphology to outstand the object and eliminate signals that we don't need. Using connected-component labeling to categorize the object, the calculated feature value of the object can be seen as the basis of the guide tile. At last, an algorithm of back propagation neural network is applied to determine the correct position of the guide tile. The result shows that the recognition rate is 87% high, suggesting that we achieve the application of Guided routes recognition and can further develop a guiding system for the visually impaired along with the effectively improved safety and convenience.
|
|---|
