CAJ | 학술논문

针对ICP算法在实际应用中存在易受异常值干扰、运算速度慢的不足,提出了一种基于K-D树的ICP改进算法。该方法通过给不同距离点对赋予不同的权值和优化K-D树建立过程中的分割策略,自动剔除迭代过程中的异常值,有效减少树的操作次数,并消除了异常值的影响。实验结果表明,该方法大大提高了ICP算法的运算速度,并改善了ICP算法的鲁棒性。
침대ICP산법재실제응용중존재역수이상치간우、운산속도만적불족,제출료일충기우K-D수적ICP개진산법。해방법통과급불동거리점대부여불동적권치화우화K-D수건립과정중적분할책략,자동척제질대과정중적이상치,유효감소수적조작차수,병소제료이상치적영향。실험결과표명,해방법대대제고료ICP산법적운산속도,병개선료ICP산법적로봉성。
Since the ICP（iterative closest point） algorithm was easily interfered by unusual values and had slow arithmetic speed in practical applications,an ICP algorithm based on K-D（k-dimensional）tree was proposed.In this method,by giving smaller weights to the points with greater distance and optimizing K-D tree during the establishment of segmentation strategy,the unusual values were automatically removed in iterative process,which could reduce the number of tree operations and eliminate the impact of outliers.The experimental result shows that the method has greatly improved the ICP algorithm operation in speed and robustness.