基于嵌入式的图像特征匹配系统设计

Design of image feature matching system based on embedded system

在计算机视觉研究领域,一个非常重要的研究项目是希望让机器可以获得像人一样观察世界的能力。因此,提出了一种将嵌入式Blackfin微处理器与融合后的图像特征匹配算法(SIST)相结合的设计方法。首先利用最大熵方法将图像清晰化,然后采用SIST算法进行图像特征点匹配,进而提高准确度。实验发现,当ratio值为0.62时,图像具有最大匹配度。实验证明本设计不仅解决了图像数据量大、难处理的问题,也在一定程度上提升了图像的匹配度,为图像处理研究提供了新思路。

In computer vision, it is a very important research purpose to let the machine gain the ability to observe the world like a human beings. This paper is based on embedded image feature matching system design. A design method combining the embedded Blackfin microprocessor with the fused image feature matching algorithm (SIST) is proposed. First the maximum entropy method is used to sharpen the image, then SIST algorithm is used to perform image feature point matching, and improve accuracy. It is found through experiments that in this system, when the ratio is 0.62, the image has the maximum matching degree. Proved by experiment, this design not only solves the problem that the amount of image data is difficult to handle, but also increase the image matching degree to some extent, providing innovative ideas for image processing research.