Abstract: Traditional-correlation-based matching methods require heavy computation time and they are sensitive to image rotation. This paper presents an image matching algorithm by combining bit planes and SIFT. It divides the two spliced images[A]and[B]into eight bit planes. The top four bit planes with visual information[A1A2A3A4]and[B1B2B3B4]are selected from the two images. Then, using the images[A1A2,][A2A3]and[A3A4]to XOR, it gets three images. Since after XOR, the image[A1A2]gets many details, but the contour is not clear;while the image[A3A4]has clear contour, but losts too many details; the image[A2A3]has clear contour, and also has enough detail information. So the paper selects the image[A2A3,]then uses image[A2A3]and the original[A]to XOR, and gets the image[A,]at the same time, uses the same method to get the image[B,]and SIFT algorithm to match again. Euclidean distance is used for image matching. Finally the RANSAC is used to process fault-tolerant, and gets a matching image. The experimental results show that the algorithm can effectively improve the matching speed, the image brightness changes, rotation and scale, with strong robustness.

Key words: bit planes, Scale Invariant Feature Transform（SIFT）, image matching, XOR

摘要： 针对传统的相关匹配算法计算量大，对图像旋转敏感等问题，提出了一种位平面和尺度不变特征变换（SIFT）相结合的图像匹配算法。将待拼接的两幅图像[A、][B]各自分解为8个位平面，对两幅图像都选择前4个具有视觉信息的位平面[A1A2A3A4]和[B1B2B3B4]；对[A1A2、][A2A3、][A3A4]图像进行异或运算，得到3幅图像。由于异或后的图像[A1A2]具有足够的细节部分，轮廓却不清晰，图像[A3A4]轮廓清晰，但是丢失了太多细节，而图像[A2A3]具有清晰的轮廓，又具有足够的细节信息，所以采用图像[A2A3]，然后与原图像[A]进行异或得到[A]，同时采用同样的方法得到图像[B]，再次采用SIFT算法进行点对匹配，利用欧氏距离进行图像匹配，最后利用RANSAC进行图像容错处理，得到一幅匹配图像。实验结果表明，该算法有效地提高了匹配速度，对图像明暗变化、尺度旋转等具有较强的健壮性。

关键词: 位平面, 尺度不变特征变换（SIFT）, 图像匹配, 异或运算