基于SIFT-SUSAN融合的草地障碍物识别算法

Grass Obstacle Recognition Algorithm Based on SIFT-SUSAN Fusion

SIFT（scale invariant feature transform）是一种对图像旋转、缩放、仿射变换具有良好不变性的机器视觉算法，在图像匹配识别上具有广泛的应用。但SIFT算法在对草地障碍物识别上存在误匹配率高和运算速度慢的问题，针对该问题提出一种SIFT-SUSAN融合算法。算法引入SUSAN算子检测并提取障碍物特征边角点，使用SUSAN提取的特征边角点和SIFT提取的特征点融合计算，对SIFT的提取特征点精简筛选后进行特征匹配。实验结果验证该算法具有可行性和有效性，提高了匹配的准确率和识别速度，且具有较好的鲁棒性。

SIFT(SCALE Invariant Feature Transform),which is a kind of machine vision algorithm with good invariance in image rotation,zoom and affine transformation,is widely used in image matching and recognition.However,there are problems of high mismatching rate and slow arithmetic speed in identification of grass obstacles by using SIFT algorithm,In ordering to solve such problems,a SIFT-SUSAN fusion algorithm is put forward.The fusion algorithm introduces the SUSAN operator to detect and extract the characteristic edge-corners of obstacles;fusion calculation is carried out by using the characteristic edge-corner points extracted with SUSAN and characteristic points extracted with SIFT;then,feature matching is conducted after downsizing and filtration of characteristic points extracted with SIFT.Experimental results verify the validity and the feasibility of the algorithm in this paper.This algorithm enhances the accuracy rate in matching and recognition speed;besides,it also has a high robustness.