Infrared image hybrid noise reduction method based on noise recognition

An infrared image and noise technology, applied in the field of infrared image hybrid noise reduction based on noise recognition, can solve problems such as image distortion

Inactive Publication Date: 2021-10-08
0 Cites 1 Cited by

AI-Extracted Technical Summary

Problems solved by technology

However, due to the influence of the acquired image through the conversion device and the environment, the image acquired i...
View more


The invention discloses an infrared image hybrid noise reduction method based on noise recognition, which comprises the following steps: firstly, carrying out 3 * 3 median filtering on a near-infrared image by using a filter to obtain an image only containing Gaussian noise, and then carrying out wavelet transform denoising processing on the image; secondly, after wave decomposition, main information of the image being almost all distributed in a low-frequency sub-band, and processing of a low-frequency part being often omitted in order to prevent important information from being damaged during denoising; performing thresholding processing on the wavelet coefficient of the high-frequency region after image decomposition, and then performing median filtering again by adopting a 5 * 5 filter; and finally, performing wavelet reconstruction by using the wavelet coefficient of each sub-band to obtain a denoised image. Through the above mode, compared with a single denoising method, the hybrid noise reduction method can more effectively remove noise and improve the signal-to-noise ratio of the near-infrared face image in view of the denoising effect of the collected night infrared image, and edge details of the image are well reserved.

Application Domain

Image enhancementImage analysis

Technology Topic

Near infra redWavelet reconstruction +7


  • Infrared image hybrid noise reduction method based on noise recognition
  • Infrared image hybrid noise reduction method based on noise recognition
  • Infrared image hybrid noise reduction method based on noise recognition


  • Experimental program(1)

Example Embodiment

[0037] DETAILED DESCRIPTION OF THE INVENTION The present invention will be described below to understand the present invention, but it should be understood, and the present invention is not limited to the scope of the specific embodiments, and in terms of ordinary skill in the art, as long as various changes Within the spirit and scope of the invention appended claims, it is apparent from the spirit and scope of the appended claims, and all inventions of the inventive concepts are all protected.
[0038] An infrared image mixing method based on noise recognition, such as figure 1 As shown, including the following steps:
[0039] S1, get noise-containing face infrared image and perform median filtering thereof;
[0040] In the present embodiment, step S1 specifically includes:
[0041] S11, constructing N × N filtering core and slides on the human face infrared image acquired by the constructed filter core, and aligns the midpoint of the built filtering core to the pixel point to be filtered;
[0042] S12, obtain the gradation value of the pixel point covered by the filter nucleus;
[0043] S13, the gradation value of the pixel point acquired by step S12 is sequentially distributed, and the median value of the order distribution is calculated;
[0044] When the interference is small, the smaller filter window should be used to remove the detail information of the image after the noise is removed; when the interference is large, the size of the filter window should be selected to achieve a better noise reduction effect. It is guaranteed that the image details are not destroyed, and the edge details are reserved. Therefore, the formula of the median value of the order distribution in step S13 is shown in the formula.
[0046] Where L k Represents the size of the square filter window, P k The proportion of interference noise points in the image line.
[0047] S14, using the pixel points covered by the midpoint of the filter wavefront;
[0048] S2, a wavelet decomposition of a human face after the median filtering in step S1 is performed to obtain a high frequency subband and a low frequency subband;
[0049] The wavelet transform is mainly completed in the image denoising process, and several important links such as wavelet decomposition of the image, the threshold process of wavelet coefficients, and the reconstruction of wavelet coefficients. The image wavelet decomposition is the selected suitable wavelet base and determine the decomposition hierarchy n, and the image is sub-wavelet decomposition. Different wavelet base functions have significant differences in different signal analysis. In the present embodiment, the small-wavelength selection generally selects the DB3 wavelet base, and the wavelet transform is a partial transformation of the space (time) and the frequency is to sufficiently highlight the features of certain aspects. Wavelet transform generates low frequency and high frequency components, and wavelet transform denoising principle is quantified to decompose the high frequency coefficient to achieve the target of the image to remove noise.
[0050] S3, the high frequency subband processes the high frequency subband after step S2 to obtain the high frequency coefficient of the noise image, and the low frequency sub-belt performs the medium value filtering process to obtain a low frequency coefficient of the noise image;
[0051] In the present embodiment, the wavelet transform denoising generally processes the high frequency components, and the high frequency coefficient W of the decomposable is threshold, and there are three main types:
[0052] Hard threshold method:
[0054] Soft threshold method:
[0056] Soft, hard-critical threshold method:
[0058] Further above in the formula, W, The wavelet coefficients after the threshold, the threshold processing, respectively, the wavelet coefficients after the threshold processing. The size of the threshold determines the effect of denoising. If the threshold is too high, an overflow occurs, and it cannot be filtered out of the noise in the image. The size of the threshold is proportional to the variance of the noise, in general, the threshold value is Or δ = (3 ~ 4) σ, σ represents the noise variance, and n is the length of the image signal.
[0059]The image after wavelet decomposing, except that the high frequency portion is scattered with a very small portion of energy, and the remaining portion of the other energy falls at the low frequency portion, and the proportion of the wavelet coefficient value of the useful signal portion is greater than the noise portion. This shows that the selection of thresholds is especially critical, which will seriously affect image quality. The absolute value is smaller than the small wave coefficient of a small threshold as a "noise", and its value is default "0", and for a larger than the threshold, it needs to be reduced. The peak signal to noise (PSNR) obtained by the hard threshold function denoising is high, but there is a phenomenon of local jitter, the peak signal to noise ratio obtained by the soft threshold function denoising is not as low as the hard threshold function, but the result is smoother. Image edge information is reserved better.
[0060] S4, the high frequency coefficient and low frequency coefficients after step S3 are fused, and refactoring is an infrared image of a noise reduction.
[0061] like figure 2 As shown, the process of near-infrared images in wavelet decomposition and reconstruction is shown. It can be seen that after wavelet decomposition, the image is a high frequency component (C) (D) (E) of the low frequency component (B) and three directions, and the detail texture of the image and the mixed noise are present in the high frequency component.
[0062] In the experiment, the noise of noise in the night infrared image. First, the noise of different strength is superimposed, and then medium-value filtering, wavelet transform deduction and mixed denoising algorithm are used to perform a noise contrast experiment on the image. Digare noise treatment results of each algorithm image 3 Indicated. The signal-to-noise ratio SNR value of the experimental results is shown in Table 1.
[0063] Table 1 signal to noise ratio of experimental results
[0065] By comparing the SNR value in Table 1, selecting a 5 * 5 medium value filter effect when selecting a DB3 wavelet group and high frequency denoising, and the SNR value reaches 22.26dB, which is significantly higher than the single wavelet transform algorithm, than 3 * 3 Medium Value Filter In binding wavelet transform, the SNR increased 2.29dB, which is more than 1.85 dB than a single use 5 * 5 medium value, which is 9.53 dB than a single wavelet change, so it is preferred to mix the denoising method when optimal. It is more than 24.2% compared to traditional 3 3 median filters, which is obvious, and optimal choice is to mix the denoising method.
[0066] The present invention is described with reference to a flowchart and / or block diagram of a method, device (system), and computer program product, in accordance with an embodiment of the present invention. It should be understood that each of the flowcharts and / or blocks in the flowchart and / or block diagram can be implemented by a computer program command, and a combination of flow and / or box in the flowchart and / or block diagram. These computer program instructions can be provided to a general purpose computer, a dedicated computer, an embedded processor, or another programmable data processing device to generate a machine such that instructions executed by the processor of the computer or other programmable data processing device. Implementation in the process Figure one Process or multiple processes and / or boxes Figure one Apparatus specified in multiple boxes or multiple boxes.
[0067] These computer program instructions can also be stored in a computer readable memory capable of booting a computer or other programmable data processing device in a particular manner, making the instructions stored in the computer readable memory generate a manufacturing article of instruction devices, which Device is implemented in the process Figure one Process or multiple processes and / or boxes Figure one The function specified in the box or multiple boxes.
[0068] These computer program instructions can also be loaded on a computer or other programmable data processing device such that a series of steps are performed on a computer or other programmable device to generate a computer implemented process, thereby executing on a computer or other programmable device. The instruction is provided for implementation Figure one Process or multiple processes and / or boxes Figure one The steps of the function specified in multiple boxes or multiple boxes.
[0069] Specific embodiments are illustrated in the present invention, the principles and embodiments of the present invention will be described, and the description of the above embodiments is intended to help understand the methods of the invention and their core ideas; at the same time, for the general articles of the art Thoughts of the invention, there will be changes in the specific embodiments and applications, which are not to be construed as limiting the invention.
[0070] One of ordinary skill in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the invention, and it is understood that the scope of the invention is not limited to such special statements and examples. One of ordinary skill in the art can make various specific modifications and combinations that are not departed from the spirit of the invention, in accordance with the disclosure of the present invention, which are still within the scope of the invention.


no PUM

Description & Claims & Application Information

We can also present the details of the Description, Claims and Application information to help users get a comprehensive understanding of the technical details of the patent, such as background art, summary of invention, brief description of drawings, description of embodiments, and other original content. On the other hand, users can also determine the specific scope of protection of the technology through the list of claims; as well as understand the changes in the life cycle of the technology with the presentation of the patent timeline. Login to view more.
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products