Power equipment multi-source image fusion method based on LatLRR and multiple filtering

Abstract

The invention discloses a power equipment multi-source image fusion method based on LatLRR and multiple filtering. The method comprises the following steps: step 1, preprocessing an input image; step 2, performing multi-filtering decomposition; step 3, carrying out LatLRR decomposition; step 4, image fusion; and step 5, fused image post-processing. Based on the LatLRR and multiple filtering technology, the image containing more details and higher contrast is generated, the state of the power equipment is detected more accurately, the infrared image and the visible light image are fused into one comprehensive image, the power equipment monitoring and fault diagnosis effect is enhanced, and the power equipment fault diagnosis efficiency is improved. And the fused image has temperature information and clear equipment details.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to a multi-source image fusion method of electric power equipment based on LatLRR and multiple filtering. Background Art

[0002] In modern power systems, ensuring the safe operation of equipment is crucial. With the continuous development of modern power systems and technological advances, infrared images and visible light images are playing an increasingly important role in the fields of power equipment monitoring and fault diagnosis. Infrared images and visible light images are the two main means of monitoring power equipment. Infrared images can reflect the temperature distribution of equipment, while visible light images can display the appearance details of equipment. Although infrared images can display temperature information, it is difficult to identify the specific equipment structure due to the lack of details and color information. Visible light images can clearly show the appearance of the equipment, but cannot reflect important information such as temperature changes;

[0003] However, traditional image fusion methods often fail to maintain image details and enhance contrast, especially in complex power equipment environments, and cannot effectively improve the monitoring effect. Therefore, the present invention proposes a multi-source image fusion method for power equipment based on LatLRR and multiple filtering to solve the problems existing in the prior art. Summary of the invention

[0004] In view of the above problems, the purpose of the present invention is to propose a multi-source image fusion method for power equipment based on LatLRR and multiple filtering. The multi-source image fusion method for power equipment based on LatLRR and multiple filtering generates images with more details and higher contrast based on LatLRR and multiple filtering technology, so as to more accurately detect the status of power equipment. By fusing infrared images and visible light images into a comprehensive image, the monitoring and fault diagnosis effects of power equipment are enhanced, so that the fused image has both temperature information and clear equipment details.

[0005] To achieve the purpose of the present invention, the present invention is implemented by the following technical scheme: a multi-source image fusion method of power equipment based on LatLRR and multiple filtering, comprising the following steps:

[0006] Step 1: Preprocess the input image, obtain the infrared image and visible light image of the power equipment, and then perform basic preprocessing such as denoising and adjusting the brightness of the image to ensure the quality of the input image;

[0007] Step 2: Multiple filtering decomposition: use multi-scale Gaussian filter and bilateral filter to decompose the infrared image and visible light image respectively. Gaussian filter is used to extract the base layer of the image, and bilateral filter is used to extract the detail layer of the image, thereby decomposing the image into a base layer, three detail layers and three fine layers;

[0008] Step 3: LatLRR decomposition. The decomposed image is input into the LatLRR model to further decompose the image from the original data, extract the global structure and local structure from the original data, and decompose X into three key parts: Z, L and E, while satisfying the formula:

[0009]

[0010] st,X=XZ+LX+E#(2);

[0011] Step 4: Image fusion: The decomposed images are divided into top layer fusion, middle layer fusion and base layer fusion to generate a fused image.

[0012] Step 5: Post-processing of fused images: Post-process the fused images to enhance image contrast and details.

[0013] A further improvement is that the base layer in step 2 mainly includes the general outline and background of the image, and the detail layer mainly includes the edges and subtle features of the image.

[0014] A further improvement is that: in step 2, I is used to represent the source image, and represents the j-th layer Gaussian filter and bilateral filter image, D j and O j represents the decomposed detail layer and fine layer, B is the basic layer (i.e. ), where the Gaussian filter parameter is selected as σ = 2.0 and the bilateral filter parameter is selected as σ s,0 =2.0,σ r,0 =2.0,σ s,j+1 =2σ s,j , σ r,j+1 =σ r,j / 2.

[0015] The further improvement is that: X, Z, L, E, XZ, and XL in step 3 represent the input original data matrix, low-rank coefficient, significant coefficient, sparse matrix, low-rank part matrix, and significant part matrix respectively; ||·|| * represents the nuclear norm, ||·||1 represents the L1 norm; λ represents the regularization parameter, which is used to control the degree of sparse representation E; st is the constraint condition.

[0016] A further improvement is that the top-level fusion in step 4, i.e., D1, O1 and I s1 The fusion method is to select the pixel point with the largest absolute value after filtering, and the top layer after LatLRR decomposition is added. The specific fusion formula is as follows:

[0017]

[0018] A further improvement is that the intermediate layer fusion in step 4, i.e. D2, O2 and I s2 Fusion, D3, O3 and I s3 Fusion, taking the pixel as a separate calculation body, the specific fusion method is as follows:

[0019]

[0020] in, λ is set to 30, σ c Set to 2, The fused intermediate layer image is represented as:

[0021]

[0022] A further improvement is that the base layer fusion in step 4, i.e. layer B and Layer fusion uses the same method as the intermediate layer fusion. The fused base layer image can be expressed as:

[0023]

[0024] in σ c Set to 4σ s,4 ;

[0025] The final fusion image of infrared and visible light images is expressed as:

[0026]

[0027] A further improvement is that the post-processing in step five includes contrast enhancement and edge sharpening of the fused image.

[0028] The beneficial effects of the present invention are as follows: the present invention generates images with more details and higher contrast based on LatLRR and multiple filtering technology, thereby achieving more accurate detection of the status of power equipment, and enhancing the monitoring and fault diagnosis effects of power equipment by fusing infrared images and visible light images into a comprehensive image, so that the fused image has both temperature information and clear equipment details. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 is a flow chart of the steps of the present invention;

[0030] Figure 2 It is the LatLRR decomposition diagram of the present invention;

[0031] Figure 3 It is a schematic diagram of decomposition of Gaussian filter and bilateral filter of the present invention;

[0032] Figure 4 is the original image of the present invention;

[0033] Figure 5 This is a diagram of the fusion result of the present invention. DETAILED DESCRIPTION

[0034] In order to deepen the understanding of the present invention, the present invention will be further described in detail below in conjunction with examples. The examples are only used to explain the present invention and do not constitute a limitation on the protection scope of the present invention.

[0035] according to Figure 1 , Figure 2 , Figure 3 , Figure 4 , Figure 5 As shown, this embodiment provides a method for multi-source image fusion of power equipment based on LatLRR and multiple filtering, comprising the following steps:

[0036] Step 1: Preprocess the input image, obtain the infrared image and visible light image of the power equipment, and then perform basic preprocessing such as denoising and adjusting the brightness of the image to ensure the quality of the input image;

[0037] Step 2: Multiple filtering decomposition: use multi-scale Gaussian filter and bilateral filter to decompose the infrared image and visible light image respectively. Gaussian filter is used to extract the base layer of the image, and bilateral filter is used to extract the detail layer of the image, thereby decomposing the image into a base layer, three detail layers and three fine layers;

[0038] The base layer mainly contains the general outline and background of the image, and the detail layer mainly contains the edges and subtle features of the image;

[0039] Let I represent the source image, and represents the j-th layer Gaussian filter and bilateral filter image, D j and O j represents the decomposed detail layer and fine layer, B is the basic layer (i.e. ), where the Gaussian filter parameter is selected as σ = 2.0 and the bilateral filter parameter is selected as σ s,0 =2.0,σ r,0 =2.0,σ s,j+1 =2σ s,j , σ r,j+1 =σ r,j / 2.

[0040] Step 3: LatLRR decomposition. The decomposed image is input into the LatLRR model to further decompose the image from the original data, extract the global structure and local structure from the original data, and decompose X into three key parts: Z, L and E, while satisfying the formula:

[0041]

[0042] st,X=XZ+LX+E#(2);

[0043] The LatLRR model decomposes the image into a low-rank part and a sparse part. The low-rank part is the background information of the image, and the sparse part is the significant features of the image and the key parts of the device.

[0044] X, Z, L, E, XZ, XL represent the input original data matrix, low-rank coefficient, significant coefficient, sparse matrix, low-rank part matrix, significant part matrix respectively; ||·|| * represents the nuclear norm, ||·||1 represents the L1 norm; λ represents the regularization parameter, which is used to control the degree of sparse representation E; st is the constraint condition;

[0045] Step 4: Image fusion: The decomposed images are divided into top layer fusion, middle layer fusion and base layer fusion to generate a fused image;

[0046] The fusion of the base layer is mainly achieved through simple average weighting, and the fusion of the detail layer adopts the sparse representation method to ensure the clarity of the details. After fusion, all layers of the image parts are reconstructed into a fused image;

[0047] Top-level fusion is D1, O1 and I s1 The fusion method is to select the pixel point with the largest absolute value after filtering, and the top layer after LatLRR decomposition is added. The specific fusion formula is as follows:

[0048]

[0049] Intermediate layer fusion: D2, O2 and I s2 Fusion, D3, O3 and I s3 Fusion, taking the pixel as a separate calculation body, the specific fusion method is as follows:

[0050]

[0051] in, λ is set to 30, σ c Set to 2, The fused intermediate layer image is represented as:

[0052]

[0053] The base layer is fused with layer B and Layer fusion uses the same method as the intermediate layer fusion. The fused base layer image can be expressed as:

[0054]

[0055] in σ c Set to 4σ s,4 ;

[0056] The final fusion image of infrared and visible light images is expressed as:

[0057]

[0058] Step 5: Post-processing the fused image: Post-processing the fused image to enhance image contrast and details;

[0059] Post-processing includes contrast enhancement and edge sharpening of the fused image to ensure that the image is clear and easy to analyze.

[0060] The fused image is compared with the images generated by other methods. First, infrared images and visible light images of different scenes are selected on the public dataset TNO to verify the effectiveness of the algorithm. The algorithms involved in the experimental comparison are: ADF, GTF, MSVD. The fusion performance is evaluated by subjective and objective methods. Then, the fusion algorithm proposed in this paper is used to fuse the infrared image and visible light image of the power equipment, and the fusion effect is demonstrated to verify its feasibility in the power field. It is concluded that this method is superior in detail preservation and contrast enhancement.

[0061] The above shows and describes the basic principles, main features and advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A method for multi-source image fusion of power equipment based on LatLRR and multiple filtering, comprising the following steps: Step 1: Preprocess the input image, obtain the infrared image and visible light image of the power equipment, and then perform basic preprocessing such as denoising and adjusting the brightness of the image to ensure the quality of the input image; Step 2: Multiple filtering decomposition: use multi-scale Gaussian filter and bilateral filter to decompose the infrared image and visible light image respectively. Gaussian filter is used to extract the base layer of the image, and bilateral filter is used to extract the detail layer of the image, thereby decomposing the image into a base layer, three detail layers and three fine layers; Step 3: LatLRR decomposition. The decomposed image is input into the LatLRR model to further decompose the image from the original data, extract the global structure and local structure, and decompose X into three key parts: Z, L, and E, while satisfying the formula: st,X=XZ+LX+E#(2); Step 4: Image fusion: The decomposed images are divided into top layer fusion, middle layer fusion and base layer fusion to generate a fused image. Step 5: Post-processing of fused images: Post-process the fused images to enhance image contrast and details.

2. The method for multi-source image fusion of power equipment based on LatLRR and multiple filtering according to claim 1 is characterized in that: The base layer in step 2 mainly includes the general outline and background of the image, and the detail layer mainly includes the edges and subtle features of the image.

3. The method for multi-source image fusion of power equipment based on LatLRR and multiple filtering according to claim 1 is characterized in that: In the step 2, I is used to represent the source image. and represents the j-th layer Gaussian filter and bilateral filter image, D j and O j represents the decomposed detail layer and fine layer, B is the basic layer (i.e. ), where the Gaussian filter parameter is selected as σ = 2.0 and the bilateral filter parameter is selected as σ s,0 =2.0,σ r,0 =2.0,σ s,j+1 =2σ s,j , σ r,j+1 =σ r,j / 2.

4. The method for multi-source image fusion of power equipment based on LatLRR and multiple filtering according to claim 1 is characterized in that: In the step 3, X, Z, L, E, XZ, and XL respectively represent the input original data matrix, low-rank coefficient, significant coefficient, sparse matrix, low-rank part matrix, and significant part matrix; ||·|| * represents the nuclear norm, ||·||1 represents the L1 norm; λ represents the regularization parameter, which is used to control the degree of sparse representation E; st is the constraint condition.

5. The method for multi-source image fusion of power equipment based on LatLRR and multiple filtering according to claim 1 is characterized in that: The top-level fusion in step 4 is D1, O1 and I s1 The fusion method is to select the pixel point with the largest absolute value after filtering, and the top layer after LatLRR decomposition is added. The specific fusion formula is as follows:

6. The method for multi-source image fusion of power equipment based on LatLRR and multiple filtering according to claim 1 is characterized in that: The intermediate layer fusion in step 4 is D2, O2 and I s2 Fusion, D3, O3 and I s3 Fusion, taking the pixel as a separate calculation body, the specific fusion method is as follows: in, λ is set to 30, σ c Set to 2, The fused intermediate layer image is represented as:

7. The method for multi-source image fusion of power equipment based on LatLRR and multiple filtering according to claim 1 is characterized by: The base layer fusion in step 4 is layer B and Layer fusion uses the same method as the intermediate layer fusion. The fused base layer image can be expressed as: in σ c Set to 4σ s,4 ; The final fusion image of infrared and visible light images is expressed as:

8. The method for multi-source image fusion of power equipment based on LatLRR and multiple filtering according to claim 1 is characterized by: The post-processing in step five includes contrast enhancement and edge sharpening of the fused image.

Images