基于导向快速与旋转简短和随机抽样一致组合的图像拼接算法优化

Image Mosaic Algorithm Optimization Based on ORB and RANSAC Combination

针对在图像拼接过程中存在图像间的特征点匹配精度低、图像拼接处存在裂缝以及图像拼接时间久的问题,提出一种基于导向快速与旋转简短(oriented fast and rotated brief, ORB)和随机抽样一致(random sample consensus, RANSAC)组合的图像拼接算法。首先,利用小波变换得到表示图像的近似、水平、垂直和对角特性的子图像分量,选取图像的近似、水平和垂直特性的子图像分量进行叠加,得到下一步进行特征提取的图像;其次,提取图像的ORB特征点并生成二进制特征描述符;再次,通过正反双向匹配对图像中的特征点进行粗匹配并使用RANSAC算法进行精度匹配;最后,利用拉普拉斯金字塔算法进行图像融合。实验结果表明:利用基于ORB和RANSAC组合的图像拼接算法对选取的图像进行提取特征平均耗时约为传统尺度不变特征转换(scale invariant feature transform, SIFT)算法的84.7%、加速鲁棒特征(speed-up robust features, SURF)算法的36.4%、ORB算法的64.9%,图像特征匹配精度提高,图像特征匹配时间缩短,图像拼接处不存在明显裂痕。基于ORB和RANSAC组合的图像拼接算法是一种优质的图像拼接算法。

To address the problems of low accuracy in matching feature points between images,cracks in the image stitching and long image stitching times in the image stitching process,an image stitching algorithm based on a combination of oriented fast and rotated brief(ORB)and random sample consensus(RANSAC)was proposed.Firstly,the wavelet transform was used to obtain sub-image components representing the approximate,horizontal,vertical and diagonal characteristics of the image,and the sub-image components of the approximate,horizontal and vertical characteristics of the image were selected and superimposed to obtain the image for the next step of feature extraction.Finally,the image was fused using the Laplace Pyramid algorithm.The experimental results show that the average time taken to extract features from the selected images using the combined ORB and RANSAC based image stitching algorithm is about 84.7%of the traditional scale invariant feature transform(SIFT)algorithm,36.4%of the speed-up robust features(SURF)algorithm and 64.9%of the ORB algorithm,the image feature matching accuracy is improved,the image feature matching time is shortened and there are no obvious cracks at the image stitching.It is concluded that the image stitching algorithm based on the combination of ORB and RANSAC is a high-quality image stitching algorithm.