Research Of High Temperature Object Edge Detection Algorithm

The measurement of high temperature object is a research hotspot and difficulty. With the development of computer vision detection technology, more and more researchers have studied in detection high temperature object with the computer vision. Edge detection is an essential step in computer vision. The effect of edge detection directly result in finally result. So in this document we have research in a good algorithm to detect high temperature object edge.Under the high temperature environment, in this document we test four types of high-temperature image: clear high-temperature image; high-temperature image with steams; high-temperature image with attachments; high-temperature image with geometric structures.Using different algorithms, the experimental results show that the classical algorithm do not do well.According to the characteristics of the high image, in this paper we combine the Nonsubsampled Contourlet Transform and phase congruency algorithm. With the advantages of the Nonsubsampled Contourlet Transform which have the ability of efficient image representation and the advantages of the phase congruency algorithm which can detect low edge, the proposed algorithm do well in edge detection.In our experiments, we design several experiments to compare these algorithms. As a result our proposed algorithm does well in edge detection of weak edge, using small pixels and extraction the whole geometrical structures.