摘要
由于智能机器人视觉系统具有非线性、时变性与强耦合性特点,引起机器人视觉避障识别率降低。且感知环境较为复杂,采用传统的方法进行障碍识别,容易发生识别错误的现象,识别效果较差。为解决上述问题,提出改进支持向量机算法的智能机器人视觉障碍识别方法。针对视觉系统采集到的原始障碍图像的噪声信息,进行中值滤波,在去除噪声的同时最大化保证了细节的完整性,依据SIFT算法理论,通过搜索空间极值点,抽取空间位置、尺度与旋转不变量,完成障碍图像的特征提取,根据改进支持向量机算法对图像特征进行分类,从而实现对视觉障碍的有效识别。实验结果表明,采用改进算法进行智能机器人视觉障碍识别,能够提高智能机器人的视觉感知能力,提高障碍识别的准确性,从而满足了智能机器人工作的实际需求。
The research of intelligent robot visual obstacle recognition method. Because of the intelligent robot vision system with nonlinear, time-varying and strong coupling characteristic, and the perception environment more complex, the traditional method is used to identify the disorder, easy to identify the error phenomenon, recognition effect is poorer. Therefore, based on improved support vector machine （SVM） algorithm of intelligent robot visual ob- stacle recognition method. For the visual system, barriers to original image noise information were collected, median filtering in removing noise of the maximum guarantee the integrity of the details at the same time, based on the theory of SIFT algorithm, through the search space is extreme value point, extract the spatial location, scale and rotation in- variant, complete obstacle image feature extraction, based on improved support vector machine （SVM） algorithm of image feature classification, so as to realize the effective recognition of the visual impairment. The experimental results show that the improved algorithm is used to .identify the intelligent robot Visual disturbances, can improve the visual perception ability of intelligent robot, improve the accuracy of obstacle recognition, and meet the actual demand of intelligent robot work.
出处
《计算机仿真》
CSCD
北大核心
2016年第1期353-356,共4页
Computer Simulation
