Space-time image flow measurement texture identification method based on frequency domain filtering technology

Abstract

The invention provides a space-time image flow measurement texture identification method based on a frequency domain filtering technology, and the method comprises the following steps: 1, reading a space-time image, and carrying out the window function processing of the space-time image; step 2, frequency spectrum construction: obtaining a spectrogram of the time-space image processed by the window function through fast two-dimensional discrete Fourier transform, center translation and amplitude spectrum calculation; 3, solving a frequency spectrum main direction through radial integration; 4,setting a threshold value and a shape of a filter according to the main direction of the frequency spectrum to perform filtering processing; and step 5, carrying out anti-center translation and two-dimensional discrete Fourier inverse transformation on the filtered space-time image frequency spectrum to obtain a noiseless space-time image texture. According to the method, noise in the space-timeimage can be effectively removed, the clear space-time image texture can be identified, the robustness, applicability and accuracy of the space-time image flow velocity measurement method are improved, and the method can adapt to monitoring of the surface flow velocity under various severe and complex water surface imaging conditions.

Description

technical field [0001] The invention belongs to the technical field of flow field measurement, and in particular relates to a texture recognition method for space-time image flow measurement based on frequency domain filtering technology. [0002] technical background [0003] Spatio-temporal image velocity measurement (STIV) is a non-contact measurement method that takes river surface images as the analysis object and obtains one-dimensional time-averaged velocity by detecting the main texture direction of the synthesized spatio-temporal image. Due to the advantages of high spatial resolution, strong real-time performance, safety and convenience, this method has special application potential in the real-time monitoring of river surface velocity and section flow. The space-time image velocity measurement method was first proposed by Professor Ichiro Fujita of Kobe University in Japan in 2007, and has successfully monitored river velocity and discharge during floods in Japan, ...

AI Technical Summary

Problems solved by technology

The above situation is impossible to avoid in practical applications, so different from the space-time images obtained in the ideal test channel, the space-time images obtained in the real environment must contain a lot of noise

The gray gradient tensor method introduces the tensor analysis theory to identify the main direction of the texture. However, the tensor analysis method is very sensitive to noise. Using this method to identify spatiotemporal images containing noise will inevitably lead to deviations or even errors.

The two-dimensional autocorrelation function method uses the autocorrelation function to identify the main direction of the texture. Similar to the previous method, the mixing of noise will cause the calculation of the autocorrelation function to deviate or even fail.

The edge spectrum analysis method uses the canny operator or grayscale transformation to detect the edge of the texture of the spatio-temporal image, and then analyzes the spectrum of the edge information to obtain the main direction of the texture. However, due to the influence of noise, the edge of the texture is usually severely deformed and lacks stability. Reliable geometric features, making the method difficult to apply

Images