Application of Fourier transform in machine vision
application of Fourier transform in machine vision
such an image1. How is it generated? 2. What does it embody? 3, how to process and to enhance the original image data?
How is such an image generated? if the Fourier transform is used to process pictures, many conditions are not considered, and the methods used are varied. On the one hand, it can filter out noise, and on the one hand, it can be used to analyze and parse the repeating region in image directly. The Fourier data obtained after processing the image is treated as image processing, which is equivalent to the spectral analysis, and the relevant data can be obtained intuitively. So the image of this spectrum analysis is very well generated in Halcon, but how does it work in OPENCV? the relevant code can achieve this effect. It's not good to Halcon. Looking for a better codeIn the end, I used this code to be able to do forward and reverse transformations.but the results need to be dealt with. Processing is quite difficult because the brightest point is connected with several lines.
second, what does such an image embody? This is a rough piece of paper on the black ink, its Fourier transform is. The first is small, then the central area is large, and finally there are several points on the edge. One proof is directionality, one is texture. After the direction transformation, we can see the main area, the transformation of the light beamThis slant picture embodies the characteristic, its highlight is not unique, proves the multidirectional nature. This reminds me of the item that the ticket recognizes. By comparison, the tilt is recognizable, but it may be difficult to identify very accurately. The estimate is possible. Especially the difference between vertical and horizontalthen I think that the two point of the light should be the size occupied by the main area. But how to do these photoelectric templates, I have no idea at present.
third, how to handle to enhance the original image? one is the direction, the other is the noise. And I can't think of a better way to repeat.
From for notes (Wiz)
Application of Fourier transform in machine vision
