Image visibility calculation method

Visibility is a very important meteorological metric and is particularly relevant in populous cities such as Hong Kong. Visibility in Hong Kong is quite low throughout the year due to air pollution from vehicle exhaust and other pollution sources in surrounding cities. By Ding. The topography and geography of the hilltops make Hong Kong’s visibility very poor when winds from the north occur. Visibility also directly affects the tourism industry, which is one of Hong Kong’s important economic pillars. Therefore, this has promoted the Hong Kong Meteorological Observatory (HKO) to provide this useful meteorological information to the public. At present, the common method for measuring visibility in the world is to use human observation to obtain it. Manual observation and determination of various land landmarks are used to determine the visible distance. This kind of investigation obviously involves a human element. In December 2006, at a workshop hosted by the City University of Hong Kong, the Hong Kong University of Science and Technology proposed the industrial application issue of “automatic calculation of atmospheric visibility based on real-time images of video cameras”, which is to seek to develop an efficient automatic It is a specialized algorithm that calculates popularity visibility by automatically processing images captured by video cameras. The key to this method is how to determine the visibility of the image by selecting the unique characteristics of the image. Star calculation, thus realizing a real-time automatic visibility calculation system. This article ll: Based on this requirement, we propose…’r a method of automatic visibility measurement and calculation based on the visual characteristics of video images. Experimental data shows that the algorithm is “jili”. Correct and valid.

The key to using fractal theory to extract marginal information in an image is to calculate the fractal dimension of the image. There are many methods to calculate the fractal dimension. Among them, the box dimension was used earlier in the field of image processing because of its relatively simple calculation and visual evaluation. Extensive, but the box dimension also has the problem that the accounting results are inconsistent with theoretical expectations. The Discrete Fractal Brown Random field (DFBR) model in fractal theory was proposed by Pentland because the grayscale images composed of surface mapping of most natural scenes satisfy the isotropic fractal Brown random field. In view of the characteristics of remote sensing impression data, the author uses the DFBR model to try to design a method for calculating the fractal dimension of pixels, and uses Matlab language programming to complete the process and extract the marginal information of high-resolution remote sensing impressions.

Use the above algorithm to extract the edges of the picture. First, confirm the size of the calculation window. When the calculation window 13 is 5×5, although the image edge information is rich, the continuity is not good, many subtle edges are mixed, and the fractal dimension changes greatly, abnormal fractal dimension values ​​appear, and road edges It is not outstanding, the plowing pattern information is redundant, and the residential area is mixed (see Figure 5b). The main reason is that when the window is 5×5, there are only two sets of variables participating in parameter fitting, resulting in low accuracy of the obtained fractal dimension. When the window is 7×7, the image outline is clear, the edge information of each object is outstanding, the abnormal fractal dimension value is small, the gullies and road boundaries in the two diagonal directions of the image are prominent, the plowing lines are clear, and the internal structure of the residential area is obvious. Marginal information is well represented (see Figure 5c).When the window El becomes larger and larger, due to the increase in the variables involved in the fitting, the fractal dimension is relatively stable. However, as the window increases, the loss of image edge information becomes more and more obvious, such as roads, gullies, boundaries, plows and Compared with the small window picture, the internal texture structure of the residential area is blurred, and the details and edge features are not obvious.

The quantitative calculation of weather visibility has extremely important practical significance for aviation, transportation and tourism. Studying a simple and practical accounting method has certain social value. The paper mainly studies a method based on image edge processing, which uses image edge gradient, contrast, brightness and other intrinsic image characteristics to quantitatively determine the visibility of preset markers on the ground based on preset thresholds, thereby achieving automatic calculation of visibility in advance. the goal of. The method in this article verifies the feasibility and effectiveness of using these intrinsic features of images to calculate visibility through comparison of experimental data.

This article is organized as follows: 1) The definition of visibility in small atmospheric science is given and the complexity of the definition of visibility in atmospheric science is analyzed, and then a simplified definition of visibility based on image processing is given, so that the basic j: The visibility algorithm for image processing is not feasible; 2) Give tt on the basis of the above definition; a visibility calculation method based on the edge of the image: 3) Use the above algorithm to give the experimental results, and analyze the accuracy of the experimental results conditions and effectiveness; 4) Conclusions and expectations.

}i At present, the common method used to measure visibility in the world is to use human observation. Manual investigators infer the actual distance that can be seen by locating various land markers (1andmark) meters. This kind of investigation obviously has artificial elements. This article will study a method based on image edge processing, which uses the inherent characteristics of images such as network image edge gradient, contrast, and brightness to determine the visibility of preset land markers based on preset thresholds, and then automatically Calculate the H that can be seen. Finally, the experimental results are given and analyzed to clarify the feasibility, reliability and accuracy of the algorithm.

Gateway de computação periférica

The so-called “ability to see” refers to the visual ability of our eyes. There are many factors that determine this visual ability. First of all, it depends on everyone’s vision. At the same time, it also determines: the observer’s field of vision depends on the shape of the target object, the size of the object, and the contrast between the background and the color characteristics of the target object and the background. Color contrast and width contrast determine the illumination during observation, whether there are brighter objects other than the target object, and more importantly, the transparency of the atmosphere. These factors will change a lot as time and place change. The change in the “visibility” value is also quite impressive.

Contactar-nos