SUPERVISION TECHNIQUE BASED TOPOGRAPHY RECOGNITION METHOD USING A VARIABLE BLOCK CAPABLE OF IMPLEMENTING THE HIGH RECOGNITION RATE OF TOPOGRAPHY BY EXTRACTING THE FEATURE VECTOR OF A BLOCK IMAGE BY THE USE OF A VARIABLE BLOCK

摘要

PURPOSE: A supervision technique based topography recognition method using a variable block is provided to obtain a high recognition rate in a minimum misclassification ratio when projecting three-dimensional information to two-dimensional information by extracting a topography feature by the usage of a variable block.;CONSTITUTION: An autonomous driving robot sets up an input topography image as multiple areas according to a distance(S10). The robot produces a block image about the input topography image by the usage of a variable block having a pixel size corresponding to each area(S30). The robot performs the Wavelet conversion of the produced block image in each YUV channel(S50). The robot extracts the feature vector of the block image(S70).;COPYRIGHT KIPO 2013;[Reference numerals] (AA) Start; (BB) End; (S10) Setting a projected 2D image as a plurality of areas; (S110) Allocating the most nearest class for a feature vector; (S30) Generating a block image for the projected 2D image by using an available block; (S50) Wavelet conversion of the block image in each YUV channel; (S70) Extracting an average value and energy as the feature vector; (S90) Artificial neural network learning of the feature vector