ISAL image quality evaluation method based on Bior and RDSIFT algorithms

Abstract

The invention provides an ISAL image quality evaluation method based on Bior and RDSIFT algorithms, which can be applied to the field of imaging detection. The method comprises the following steps: processing an ISAL laser inverse synthetic aperture image by using a biorthogonal wavelet basis algorithm (Bior) to obtain time-frequency domain features; processing position information of pixel points in the ISAL laser inverse synthetic aperture image by using a feature transformation algorithm (RDSIFT) to obtain directional features; carrying out convolution on the gray features and the directivity features of the ISAL laser inverse synthetic aperture image to obtain scale invariant features; performing gradient transformation on the scale invariant features to obtain texture features; and fusing and inputting the time-frequency domain features and the texture features into a support vector machine model, and outputting an image quality evaluation result.

Description

Technical Field

[0001] The present disclosure relates to the field of imaging detection, and more specifically, to an ISAL image quality assessment method based on Bior and RDSIFT algorithms. Background Art

[0002] With the continuous maturity and improvement of laser inverse synthetic aperture imaging technology, various imaging algorithms have emerged, and many laser inverse synthetic aperture imaging systems have been put into use. The imaging quality of imaging algorithms has received much attention, and a more objective method is needed to evaluate image quality, rather than relying solely on the subjective evaluation of experts. Due to the complex texture and rich detail features of ISAL laser inverse synthetic aperture images, the evaluation index methods of related technologies, such as equivalent visual number and average gradient, are used to evaluate the quality of images. The evaluation results are not accurate and lack quality evaluation of image characteristics. Existing evaluation algorithms are difficult to fully capture complex local, texture, and edge features, which will result in the loss of important features and low image evaluation accuracy. Summary of the invention

[0003] In view of this, the present disclosure provides an ISAL image quality assessment method based on Bior and RDSIFT algorithms.

[0004] One aspect of the present disclosure provides an ISAL image quality assessment method based on Bior and RDSIFT algorithms, comprising: using a biorthogonal wavelet basis algorithm (Bior) to process an ISAL laser inverse synthetic aperture image to obtain time-frequency domain features; using a feature transformation algorithm (RDSIFT) to process the position information of pixel points in the ISAL laser inverse synthetic aperture image to obtain directional features; convolving the grayscale features and directional features of the ISAL laser inverse synthetic aperture image to obtain scale-invariant features; performing gradient transformation on the scale-invariant features to obtain texture features; fusing the time-frequency domain features and the texture features and inputting them into a support vector machine model to output an image quality assessment result.

[0005] According to an embodiment of the present disclosure, an ISAL laser inverse synthetic aperture image is processed using a feature transformation algorithm (RDSIFT) to obtain directional features, including: processing position information of pixel points in the ISAL laser inverse synthetic aperture image according to a Gaussian function to obtain spatial features; processing position information and spatial features according to a radial basis function to obtain radial basis features; processing position information and radial basis features according to a directional filtering algorithm to obtain directional features.

[0006] According to an embodiment of the present disclosure, the ISAL laser inverse synthetic aperture image is processed by using a biorthogonal wavelet basis algorithm (Bior) to obtain time-frequency domain features, including: transforming the ISAL laser inverse synthetic aperture image based on the biorthogonal wavelet basis algorithm (Bior) to obtain a first reference image, a first horizontal azimuth image, a first vertical azimuth image, and a first diagonal azimuth image, wherein the first reference image represents an image whose image grayscale value change frequency is less than a first threshold value, and the first horizontal azimuth image, the first vertical azimuth image, and the first diagonal azimuth image all represent images whose image grayscale value change frequency is greater than a second threshold value; transforming the first reference image based on the biorthogonal wavelet basis algorithm (Bior) to obtain a second reference image, a second horizontal azimuth image, a second vertical azimuth image, and a second diagonal azimuth image, wherein the second reference image represents an image whose image grayscale value change frequency is less than the first threshold value, and the second horizontal azimuth image, the first vertical azimuth image, and the first diagonal azimuth image all represent images whose image grayscale value change frequency is greater than a second threshold value. The second horizontal azimuth image, the second vertical azimuth image and the second diagonal azimuth image all represent images whose grayscale value change frequency is greater than a second threshold value; the second reference image, the second horizontal azimuth image, the second vertical azimuth image, the second diagonal azimuth image, the first horizontal azimuth image, the first vertical azimuth image and the first diagonal azimuth image are quantized respectively to obtain the second reference quantized image, the second horizontal azimuth quantized image, the second vertical azimuth quantized image, the second diagonal azimuth quantized image, the first horizontal azimuth quantized image, the first vertical azimuth quantized image and the first diagonal azimuth quantized image; feature extraction is performed on the second reference quantized image, the second horizontal azimuth quantized image, the second vertical azimuth quantized image, the second diagonal azimuth quantized image, the first horizontal azimuth quantized image, the first vertical azimuth quantized image and the first diagonal azimuth quantized image to obtain the time-frequency domain features corresponding to the ISAL laser inverse synthetic aperture image.

[0007] According to an embodiment of the present disclosure, a second reference image, a second horizontal orientation image, a second vertical orientation image, a second diagonal orientation image, a first horizontal orientation image, a first vertical orientation image, and a first diagonal orientation image are quantized to obtain a second reference quantized image, a second horizontal orientation quantized image, a second vertical orientation quantized image, a second diagonal orientation quantized image, a first horizontal orientation quantized image, a first vertical orientation quantized image, and a first diagonal orientation quantized image. The quantization process includes: processing pixel wavelet coefficients of the first diagonal orientation image according to a noise standard deviation algorithm to obtain a first noise standard deviation; processing the first noise standard deviation and the number of pixels of the first diagonal orientation image according to a threshold algorithm to obtain a first threshold; and based on the first threshold, processing the first noise standard deviation and the number of pixels of the first diagonal orientation image. A horizontal azimuth image, a vertical azimuth image, and a first diagonal azimuth image are quantized to obtain a first horizontal azimuth quantized image, a first vertical azimuth quantized image, and a first diagonal azimuth quantized image; pixel wavelet coefficients of the second diagonal azimuth image are processed according to a noise standard deviation algorithm to obtain a second noise standard deviation; the second noise standard deviation and the number of pixels of the second diagonal azimuth image are processed according to a threshold algorithm to obtain a second threshold; based on the second threshold, pixel wavelet coefficients of the second reference image, the second horizontal azimuth image, the second vertical azimuth image, and the second diagonal azimuth image are quantized to obtain a second reference quantized image, a second horizontal azimuth quantized image, a second vertical azimuth quantized image, and a second diagonal quantized image.

[0008] According to an embodiment of the present disclosure, feature extraction is performed on the second reference quantized image, the second horizontal azimuth quantized image, the second vertical azimuth quantized image, the second diagonal azimuth quantized image, the first horizontal azimuth quantized image, the first vertical azimuth quantized image, and the first diagonal azimuth quantized image to obtain time-frequency domain features corresponding to the ISAL laser inverse synthetic aperture image, including: time-frequency domain feature extraction is performed on the second reference quantized image, the second horizontal azimuth quantized image, the second vertical azimuth quantized image, the second diagonal azimuth quantized image, the first horizontal azimuth quantized image, the first vertical azimuth quantized image, and the first diagonal azimuth quantized image, respectively, to obtain multiple time-frequency domain sub-features, wherein the time-frequency domain sub-features include frequency band entropy information, frequency band energy information, frequency band average information, and frequency band standard deviation information; multiple time-frequency domain sub-features are fused to obtain the time-frequency domain features corresponding to the ISAL laser inverse synthetic aperture image.

[0009] According to an embodiment of the present disclosure, gradient transformation is performed based on scale-invariant features to obtain texture features, including: dividing an ISAL laser inverse synthetic aperture image into regions to obtain a plurality of local regions; for an i-th local region among the plurality of local regions, processing the scale-invariant features corresponding to each pixel point located in the i-th local region using a gradient algorithm to obtain a gradient magnitude and a gradient direction of each pixel point; determining an i-th gradient histogram corresponding to the i-th local region based on the gradient magnitude and the gradient direction of each pixel point; encoding the i-th local region according to the i-th gradient histogram to obtain an i-th local texture feature; and determining a texture feature according to the local texture features corresponding to each of the plurality of local regions.

[0010] According to an embodiment of the present disclosure, a support vector machine model is trained based on the following operations: obtaining training samples, wherein the training samples include sample ISAL laser inverse synthetic aperture images and classification labels, the classification labels characterizing quality assessment categories of the sample ISAL laser inverse synthetic aperture images, the quality assessment categories including shooting posture categories and clarity categories; performing feature extraction on the sample ISAL laser inverse synthetic aperture images to obtain sample time-frequency domain features and sample texture features; training a support vector machine model based on the sample time-frequency domain features, the sample texture features, and the classification labels to obtain a trained support vector machine model.

[0011] According to an embodiment of the present disclosure, a support vector machine model is trained based on sample time-frequency domain features, sample texture features, and classification labels, and the trained support vector machine model includes: inputting sample time-frequency domain features and sample texture features into the support vector machine model, and outputting sample image quality assessment results; using a loss function to process the sample image quality assessment results and classification labels to obtain a loss value; and training the support vector machine model based on the loss value to obtain a trained support vector machine model.

[0012] Another aspect of the present disclosure provides an ISAL image quality assessment device, comprising:

[0013] The first processing module is used to process the ISAL laser inverse synthetic aperture image using a biorthogonal wavelet basis algorithm (Bior) to obtain time-frequency domain features;

[0014] The second processing module is used to process the position information of the pixel points in the ISAL laser inverse synthetic aperture image by using a feature transformation algorithm (RDSIFT) to obtain a directional feature;

[0015] The convolution module is used to convolve the grayscale features and directional features of the ISAL laser inverse synthetic aperture image to obtain scale-invariant features;

[0016] Transformation module, used to perform gradient transformation on scale-invariant features to obtain texture features;

[0017] The evaluation module is used to fuse the time-frequency domain features and texture features into the support vector machine model and output the image quality evaluation results.

[0018] Another aspect of the present disclosure provides an electronic device, comprising:

[0019] one or more processors;

[0020] a memory for storing one or more programs,

[0021] When one or more programs are executed by one or more processors, the one or more processors implement the ISAL image quality assessment method based on the Bior and RDSIFT algorithms as described above.

[0022] According to the ISAL image quality assessment method based on Bior and RDSIFT algorithms provided by the present disclosure, the ISAL laser inverse synthetic aperture image is processed by using a biorthogonal wavelet basis algorithm (Bior) to obtain time-frequency domain features; a feature transformation algorithm (RDSIFT) is used to obtain directional features; grayscale features and directional features are convolved to obtain scale-invariant features; scale-invariant features are gradient transformed to obtain texture features; time-frequency domain features and texture features are fused and input into a support vector machine model to output an image quality assessment result. Since the biorthogonal wavelet basis algorithm (Bior) is used to extract time-frequency domain features, and the feature transformation algorithm (RDSIFT) is used to further extract directional features, scale-invariant features, and texture features, not only can the scale-invariant features be detected, but also the extraction capabilities of texture and directional features are increased, and the processing efficiency is improved. In addition, the fusion of multiple features realizes a comprehensive assessment of the ISAL laser inverse synthetic aperture image, thereby improving the accuracy of image quality assessment. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] The above and other objects, features and advantages of the present disclosure will become more apparent through the following description of the embodiments of the present disclosure with reference to the accompanying drawings, in which:

[0024] Figure 1 A flowchart of an ISAL image quality assessment method based on Bior and RDSIFT algorithms according to an embodiment of the present disclosure is shown;

[0025] Figure 2 An example schematic diagram of transforming an ISAL laser inverse synthetic aperture image according to an embodiment of the present disclosure is shown;

[0026] Figure 3 A flowchart of training a support vector machine model according to an embodiment of the present disclosure is shown;

[0027] Figure 4A block diagram of an ISAL image quality assessment device according to an embodiment of the present disclosure is shown; and

[0028] Figure 5 A block diagram of an electronic device suitable for implementing the ISAL image quality assessment method based on the Bior and RDSIFT algorithms according to an embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0029] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present disclosure. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, it is apparent that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.

[0030] The terms used herein are only for describing specific embodiments and are not intended to limit the present disclosure. The terms "comprise", "include", etc. used herein indicate the existence of features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.

[0031] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.

[0032] When using expressions such as "at least one of A, B, and C, etc.", they should generally be interpreted according to the meaning of the expression commonly understood by those skilled in the art (for example, "a system having at least one of A, B, and C" should include but is not limited to a system having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.).

[0033] In the process of realizing the present disclosure, it is found that with the continuous maturity and improvement of laser inverse synthetic aperture imaging technology, various imaging algorithms continue to emerge, and many laser inverse synthetic aperture imaging systems have been put into use. The imaging quality of the imaging algorithm has received much attention, and a more objective method is needed to evaluate the image quality, rather than relying solely on the subjective evaluation of experts. Due to the complex texture and rich detail features of ISAL laser inverse synthetic aperture images, the evaluation index methods of related technologies, such as equivalent visual number, average gradient, etc., are used to evaluate the quality of the image. The evaluation results are not accurate, and there is a lack of quality evaluation of image characteristics. Existing evaluation algorithms are difficult to fully capture complex local, texture, and edge features, which will result in the loss of important features and low image evaluation accuracy.

[0034] In view of this, the embodiment of the present disclosure provides an ISAL image quality assessment method based on Bior and RDSIFT algorithms. The method includes: using a biorthogonal wavelet basis algorithm (Bior) to process the ISAL laser inverse synthetic aperture image to obtain time-frequency domain features; using a feature transformation algorithm (RDSIFT) to process the position information of the pixel points in the ISAL laser inverse synthetic aperture image to obtain directional features; convolving the grayscale features and directional features of the ISAL laser inverse synthetic aperture image to obtain scale-invariant features; performing gradient transformation on the scale-invariant features to obtain texture features; fusing the time-frequency domain features and texture features into a support vector machine model to output an image quality assessment result.

[0035] In the technical solution of the present disclosure, the user information (including but not limited to user personal information, user image information, user device information, such as location information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved are all information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of the relevant data comply with relevant laws, regulations and standards, take necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0036] Figure 1 A flowchart of an ISAL image quality assessment method based on Bior and RDSIFT algorithms according to an embodiment of the present disclosure is shown.

[0037] like Figure 1 As shown, the method 100 includes operations S110 to S150.

[0038] In operation S110, the ISAL laser inverse synthetic aperture image is processed using a biorthogonal wavelet basis algorithm (Bior) to obtain time-frequency domain features.

[0039] According to an embodiment of the present disclosure, the biorthogonal wavelet basis algorithm may be constructed based on a wavelet function (Biorthogonal, bior).

[0040] According to an embodiment of the present disclosure, a laser inverse synthetic aperture image is obtained by imaging between a radar inverse synthetic aperture laser radar (ISAL) and a target object.

[0041] According to an embodiment of the present disclosure, the time-frequency domain features characterize the frequency fluctuation and variation information of the ISAL laser inverse synthetic aperture image.

[0042] In operation S120, the position information of the pixel points in the ISAL laser inverse synthetic aperture image is processed using a feature transformation algorithm (RDSIFT) to obtain a directional feature.

[0043] According to an embodiment of the present disclosure, the feature transformation algorithm may be constructed based on a radial basis function network direction scale-invariant feature transform (RDSIFT).

[0044] According to an embodiment of the present disclosure, the directional feature characterizes the multi-directional filtering change information of the ISAL laser inverse synthetic aperture image.

[0045] In operation S130, convolution is performed on the grayscale features and the directional features of the ISAL laser inverse synthetic aperture image to obtain scale-invariant features.

[0046] According to an embodiment of the present disclosure, the grayscale information of the ISAL laser inverse synthetic aperture image is extracted to obtain grayscale features. The grayscale features and directional features of the ISAL laser inverse synthetic aperture image are convolved to obtain scale-invariant features.

[0047] According to an embodiment of the present disclosure, the scale-invariant feature is invariant to the scale and rotation of the ISAL laser inverse synthetic aperture image. The ISAL laser inverse synthetic aperture image is processed with different scales and rotations, and the scale-invariant feature extracted from the ISAL laser inverse synthetic aperture image after the processing with different scales and rotations is invariant.

[0048] In one embodiment, the scale-invariant feature The calculation is shown in formula (1):

[0049] (1)

[0050] in, Characterize the grayscale features, Characterize directional characteristics, Characterizes the filter direction, Characterizes the phase shift, Characterizes the standard deviation of pixel coordinates, Represents the horizontal coordinate of the pixel in the ISAL laser inverse synthetic aperture image, Represents the vertical coordinate of a pixel in the ISAL laser inverse synthetic aperture image.

[0051] In operation S140, gradient transformation is performed on the scale-invariant feature to obtain a texture feature.

[0052] According to the embodiments of the present disclosure, each key pixel point can be located based on scale-invariant features, such as the size sub-features and principal curvature sub-features of the pixel points in the ISAL laser inverse synthetic aperture image. The gradient direction distribution of adjacent pixel points is used as the specified direction parameter. After calculating the magnitude and direction of the gradient between each key pixel point and the adjacent pixel points, a histogram is constructed to extract texture features.

[0053] According to an embodiment of the present disclosure, texture features characterize the property information of the surface structure and tissue arrangement of the ISAL laser inverse synthetic aperture image.

[0054] In operation S150, the time-frequency domain features and the texture features are fused and input into a support vector machine model, and an image quality assessment result is output.

[0055] According to an embodiment of the present disclosure, the time-frequency domain features and the texture features are fused, and the fused features are input into a support vector machine (SVM) model to output an image quality assessment result.

[0056] According to an embodiment of the present disclosure, the image quality assessment result represents a prediction result of assessing the ISAL laser inverse synthetic aperture image quality.

[0057] For example, the ISAL laser inverse synthetic aperture image quality categories can include posture categories and clarity categories. The posture categories are divided into six categories: front view, side view, top view, bottom view, rear view and squint. The clarity categories are divided into four categories: excellent, good, medium and poor. The support vector machine model is a binary classifier, and the support vector machine model outputs category probabilities or scores.

[0058] According to the embodiments of the present disclosure, since the time-frequency domain features are extracted by using the biorthogonal wavelet basis algorithm (Bior), and the directional features, scale-invariant features, and texture features are further extracted by using the feature transformation algorithm (RDSIFT), not only the scale-invariant features can be detected, but also the extraction capability of texture and directional features is increased, thereby improving the processing efficiency. In addition, the fusion of multiple features realizes the comprehensive evaluation of the ISAL laser inverse synthetic aperture image, thereby improving the accuracy of image quality evaluation.

[0059] According to an embodiment of the present disclosure, an ISAL laser inverse synthetic aperture image is processed using a feature transformation algorithm (RDSIFT) to obtain directional features, including: processing position information of pixel points in the ISAL laser inverse synthetic aperture image according to a Gaussian function to obtain spatial features; processing position information and spatial features according to a radial basis function to obtain radial basis features; processing position information and radial basis features according to a directional filtering algorithm to obtain directional features.

[0060] According to an embodiment of the present disclosure, a feature transformation algorithm (RDSIFT) is constructed based on a Gaussian function, a radial basis function and a directional filtering algorithm.

[0061] According to an embodiment of the present disclosure, the Gaussian function may be a two-dimensional Gaussian function, and the spatial feature represents the spatial position state information.

[0062] In one embodiment, the Gaussian function As shown in formula (2):

[0063] (2)

[0064] in, Characterizes the standard deviation of pixel coordinates, Represents the horizontal coordinate of the pixel in the ISAL laser inverse synthetic aperture image, Represents the vertical coordinate of a pixel in the ISAL laser inverse synthetic aperture image.

[0065] According to an embodiment of the present disclosure, the pixel coordinate standard deviation is used to control the scale of blur.

[0066] According to an embodiment of the present disclosure, a radial basis function network (RBF) is used to process the situation where the function value from the center point varies with the distance, thereby adjusting the response of the Gaussian kernel in the Gaussian function, and is more sensitive to capturing specific types of image features.

[0067] In one embodiment, the directional characteristics The calculation is shown in formula (3):

[0068] (3)

[0069] in, Represents the standard deviation in the horizontal direction, Characterizes the standard deviation in the vertical direction.

[0070] According to an embodiment of the present disclosure, the respective standard deviations in the horizontal direction and the vertical direction can control the scale of blur in different directions.

[0071] According to an embodiment of the present disclosure, radial basis features characterize deep-level, highly sensitive directional features.

[0072] According to an embodiment of the present disclosure, the directional filtering algorithm is a direction-sensitive multi-scale filter, which is used to improve the detection capability of specific directional features.

[0073] In one embodiment, the directional filtering algorithm As shown in formula (4):

[0074] (4)

[0075] in, , , Characterize the filter wavelength, Characterizes the direction of the filter, Characterizes phase shift.

[0076] According to an embodiment of the present disclosure, the filter wavelength is used to control the frequency of the filter.

[0077] According to an embodiment of the present disclosure, in different training processes, by changing the values ​​of σ, β, γ and θ, a set of directional features of different scales and directions are generated. The directional features characterize the filtering change information of the ISAL laser inverse synthetic aperture image at different scales and directions.

[0078] According to an embodiment of the present disclosure, a radial basis function is combined with a directional filtering algorithm to replace Gaussian blur, and a radial basis directional scale space is established, thereby enhancing the detection capability of local features of the ISAL laser inverse synthetic aperture image while maintaining scale invariance.

[0079] According to an embodiment of the present disclosure, the ISAL laser inverse synthetic aperture image is processed using a biorthogonal wavelet basis algorithm (Bior), and the time-frequency domain features obtained include:

[0080] Based on the biorthogonal wavelet basis algorithm (Bior), the ISAL laser inverse synthetic aperture image is transformed to obtain a first reference image, a first horizontal azimuth image, a first vertical azimuth image, and a first diagonal azimuth image, wherein the first reference image represents an image whose image grayscale value change frequency is less than a first threshold value, and the first horizontal azimuth image, the first vertical azimuth image, and the first diagonal azimuth image all represent images whose image grayscale value change frequency is greater than a second threshold value; based on the biorthogonal wavelet basis algorithm (Bior), the first reference image is transformed to obtain a second reference image, a second horizontal azimuth image, a second vertical azimuth image, and a second diagonal azimuth image, wherein the second reference image represents an image whose image grayscale value change frequency is less than the first threshold value, and the second horizontal azimuth image, the second vertical azimuth image, and the second diagonal azimuth image all represent An image whose grayscale value change frequency is greater than a second threshold value; quantizing the second reference image, the second horizontal azimuth image, the second vertical azimuth image, the second diagonal azimuth image, the first horizontal azimuth image, the first vertical azimuth image, and the first diagonal azimuth image respectively to obtain the second reference quantized image, the second horizontal azimuth quantized image, the second vertical azimuth quantized image, the second diagonal azimuth quantized image, the first horizontal azimuth quantized image, the first vertical azimuth quantized image, and the first diagonal azimuth quantized image; extracting features from the second reference quantized image, the second horizontal azimuth quantized image, the second vertical azimuth quantized image, the second diagonal azimuth quantized image, the first horizontal azimuth quantized image, the first vertical azimuth quantized image, and the first diagonal azimuth quantized image to obtain the time-frequency domain features corresponding to the ISAL laser inverse synthetic aperture image.

[0081] According to an embodiment of the present disclosure, the ISAL laser inverse synthetic aperture image is decomposed and transformed twice using a biorthogonal wavelet basis algorithm (Bior), wherein the first-level decomposition transformation obtains a first reference image, a first horizontal azimuth image, a first vertical azimuth image, and a first diagonal azimuth image; the first-level decomposition transformation obtains a first reference image, a first horizontal azimuth image, a first vertical azimuth image, and a first diagonal azimuth image; and the second-level decomposition transformation obtains a second reference image, a second horizontal azimuth image, a second vertical azimuth image, and a second diagonal azimuth image.

[0082] According to an embodiment of the present disclosure, the size of the ISAL laser inverse synthetic aperture image is 1024*1024, and the ISAL laser inverse synthetic aperture image is subjected to a first-level decomposition transformation using a biorthogonal wavelet basis algorithm (Bior) to obtain a first reference image, a first horizontal azimuth image, a first vertical azimuth image, and a first diagonal azimuth image, all of which have an image size of 512*512.

[0083] According to an embodiment of the present disclosure, the first threshold is used to define the low-frequency and high-frequency intervals. The first reference image represents an image whose grayscale value variation frequency is less than the first threshold, and the first reference image is a low-frequency image.

[0084] According to an embodiment of the present disclosure, the first horizontal orientation image, the first vertical orientation image, and the first diagonal orientation image all represent images whose grayscale value change frequency is greater than the second threshold value, and the first horizontal orientation image, the first vertical orientation image, and the first diagonal orientation image are all high-frequency images. Taking the first reference image as a reference image, the first horizontal orientation image is located in the horizontal direction of the first reference image, and the first vertical orientation image is located in the vertical direction of the first reference image. The first diagonal orientation image is located in the diagonal direction of the first reference image.

[0085] According to an embodiment of the present disclosure, the size of the first reference image is 256*256, and the first reference image is subjected to a two-level decomposition transformation using a biorthogonal wavelet basis algorithm (Bior) to obtain a second reference image, a second horizontal orientation image, a second vertical orientation image, and a second diagonal orientation image, all of which have image sizes of 256*256.

[0086] According to an embodiment of the present disclosure, the first threshold is used to define the low-frequency and high-frequency intervals. The second reference image represents an image whose grayscale value variation frequency is less than the first threshold, and the second reference image is a low-frequency image.

[0087] According to an embodiment of the present disclosure, the second horizontal orientation image, the second vertical orientation image, and the second diagonal orientation image all represent images whose grayscale value change frequency is greater than the second threshold value, and the second horizontal orientation image, the second vertical orientation image, and the second diagonal orientation image are all high-frequency images. Taking the second reference image as a reference image, the second horizontal orientation image is located in the horizontal direction of the second reference image, and the second vertical orientation image is located in the vertical direction of the second reference image. The second diagonal orientation image is located in the diagonal direction of the second reference image.

[0088] According to an embodiment of the present disclosure, the wavelet coefficients of the second reference image, the second horizontal azimuth image, the second vertical azimuth image, the second diagonal azimuth image, the first horizontal azimuth image, the first vertical azimuth image, and the first diagonal azimuth image are quantized to obtain the second reference quantized image, the second horizontal azimuth quantized image, the second vertical azimuth quantized image, the second diagonal azimuth quantized image, the first horizontal azimuth quantized image, the first vertical azimuth quantized image, and the first diagonal azimuth quantized image.

[0089] According to the embodiments of the present disclosure, time-frequency domain features are extracted from the second reference quantized image, the second horizontal azimuth quantized image, the second vertical azimuth quantized image, the second diagonal azimuth quantized image, the first horizontal azimuth quantized image, the first vertical azimuth quantized image, and the first diagonal azimuth quantized image, respectively, to obtain multiple sub-time-frequency domain features, and the multiple sub-time-frequency domains are merged to obtain the time-frequency domain features corresponding to the ISAL laser inverse synthetic aperture image.

[0090] Figure 2An example schematic diagram of transforming an ISAL laser inverse synthetic aperture image according to an embodiment of the present disclosure is shown.

[0091] like Figure 2 As shown, after the ISAL laser inverse synthetic aperture image is decomposed twice using the biorthogonal wavelet basis algorithm (Bior), the first horizontal azimuth image HL1, the first vertical azimuth image LH1, the first diagonal azimuth image HH1, the second reference image LL2, the second horizontal azimuth image HL2, the second vertical azimuth image LH2, and the second diagonal azimuth image HH2 are finally obtained.

[0092] According to an embodiment of the present disclosure, quantizing the second reference image, the second horizontal azimuth image, the second vertical azimuth image, the second diagonal azimuth image, the first horizontal azimuth image, the first vertical azimuth image, and the first diagonal azimuth image respectively to obtain the second reference quantized image, the second horizontal azimuth quantized image, the second vertical azimuth quantized image, the second diagonal azimuth quantized image, the first horizontal azimuth quantized image, the first vertical azimuth quantized image, and the first diagonal azimuth quantized image includes:

[0093] Process the pixel wavelet coefficients of the first diagonal orientation image according to the noise standard deviation algorithm to obtain the first noise standard deviation; process the first noise standard deviation and the number of pixels of the first diagonal orientation image according to the threshold algorithm to obtain the first threshold; based on the first threshold, quantize the first horizontal orientation image, the first vertical orientation image, and the first diagonal orientation image to obtain the first horizontal orientation quantized image, the first vertical orientation quantized image, and the first diagonal orientation quantized image; process the pixel wavelet coefficients of the second diagonal orientation image according to the noise standard deviation algorithm to obtain the second noise standard deviation; process the second noise standard deviation and the number of pixels of the second diagonal orientation image according to the threshold algorithm to obtain the second threshold; based on the second threshold, quantize the pixel wavelet coefficients of the second reference image, the second horizontal orientation image, the second vertical orientation image, and the second diagonal orientation image to obtain the second reference quantized image, the second horizontal orientation quantized image, the second vertical orientation quantized image, and the second diagonal orientation quantized image.

[0094] According to the embodiments of the present disclosure, the ISAL laser inverse synthetic aperture image needs to undergo a two-layer decomposition transformation. For the image in each layer of the decomposition, the wavelet coefficients of the pixels in the image are threshold quantized according to the threshold corresponding to each layer.

[0095] In one embodiment, the noise standard deviation algorithm is shown in formula (5):

[0096] (5)

[0097] in, Characterize the decomposition layer, Characterize the pixel wavelet coefficients of the first diagonal image, Characterize the pixel wavelet coefficients of the second diagonal image, Characterize the median function, Characterizes the first noise standard deviation, Characterizes the second noise standard deviation.

[0098] According to an embodiment of the present disclosure, the first noise standard variance is obtained according to the first noise standard deviation; and the second noise standard variance is obtained according to the second noise standard deviation.

[0099] According to an embodiment of the present disclosure, a threshold algorithm may be constructed based on a general threshold function.

[0100] In one embodiment, the threshold algorithm is shown in formula (6):

[0101] (6)

[0102] in, Characterize the decomposition layer, Characterize the first threshold, Characterize the second threshold, Characterization The total number of image pixels corresponding to the layer, is 1024*1024, is 512*512, Characterization The standard deviation of the noise corresponding to the layer, Characterizes the first noise standard deviation, Characterizes the second noise standard deviation.

[0103] According to an embodiment of the present disclosure, based on a first threshold, a soft threshold processing algorithm is used to quantize the wavelet coefficients of pixel points in an image to obtain a first horizontal quantized image, a first vertical quantized image, and a first diagonal quantized image; based on a second threshold, a soft threshold processing algorithm is used to quantize the wavelet coefficients of pixel points in an image to obtain a second baseline quantized image, a second horizontal quantized image, a second vertical quantized image, and a second diagonal quantized image.

[0104] In one embodiment, the soft threshold processing algorithm is shown in formula (7):

[0105] (7)

[0106] in, Characterization The wavelet coefficients of the image pixels corresponding to the layer, including pixel wavelet coefficients of the first horizontal orientation image, the first vertical orientation image, and the first diagonal orientation image, including pixel wavelet coefficients of the second reference image, the second horizontal orientation image, the second vertical orientation image, and the second diagonal orientation image, Characterize the first threshold, Characterize the second threshold, including pixel point wavelet coefficients of the first horizontal quantized image, the first vertical quantized image, and the first diagonal quantized image, It includes pixel point wavelet coefficients of the second reference quantized image, the second horizontal quantized image, the second vertical quantized image, and the second diagonal quantized image.

[0107] According to an embodiment of the present disclosure, feature extraction is performed on the second reference quantized image, the second horizontal azimuth quantized image, the second vertical azimuth quantized image, the second diagonal azimuth quantized image, the first horizontal azimuth quantized image, the first vertical azimuth quantized image, and the first diagonal azimuth quantized image to obtain time-frequency domain features corresponding to the ISAL laser inverse synthetic aperture image, including: time-frequency domain feature extraction is performed on the second reference quantized image, the second horizontal azimuth quantized image, the second vertical azimuth quantized image, the second diagonal azimuth quantized image, the first horizontal azimuth quantized image, the first vertical azimuth quantized image, and the first diagonal azimuth quantized image, respectively, to obtain multiple time-frequency domain sub-features, wherein the time-frequency domain sub-features include frequency band entropy information, frequency band energy information, frequency band average information, and frequency band standard deviation information; multiple time-frequency domain sub-features are fused to obtain the time-frequency domain features corresponding to the ISAL laser inverse synthetic aperture image.

[0108] According to an embodiment of the present disclosure, the frequency band entropy information, frequency band energy information, frequency band average information and frequency band standard deviation information are performed on the second reference quantized image to obtain the time-frequency domain sub-features corresponding to the second reference quantized image.

[0109] According to an embodiment of the present disclosure, the frequency band entropy information, frequency band energy information, frequency band average information and frequency band standard deviation information are performed on the second horizontal quantized image to obtain the time-frequency domain sub-features corresponding to the second horizontal quantized image.

[0110] According to an embodiment of the present disclosure, the frequency band entropy information, frequency band energy information, frequency band average information and frequency band standard deviation information are performed on the second vertical quantized image to obtain the time-frequency domain sub-features corresponding to the second vertical quantized image.

[0111] According to an embodiment of the present disclosure, the frequency band entropy information, frequency band energy information, frequency band average information and frequency band standard deviation information are performed on the second diagonal quantized image to obtain the time-frequency domain sub-features corresponding to the second diagonal quantized image.

[0112] According to an embodiment of the present disclosure, the frequency band entropy information, frequency band energy information, frequency band average information and frequency band standard deviation information are performed on the first diagonal quantized image to obtain the time-frequency domain sub-features corresponding to the first diagonal quantized image.

[0113] According to an embodiment of the present disclosure, the frequency band entropy information, frequency band energy information, frequency band average information and frequency band standard deviation information are performed on the first vertical quantized image to obtain the time-frequency domain sub-features corresponding to the first vertical quantized image.

[0114] According to an embodiment of the present disclosure, the frequency band entropy information, frequency band energy information, frequency band average information and frequency band standard deviation information are performed on the first horizontal quantized image to obtain the time-frequency domain sub-features corresponding to the first horizontal quantized image.

[0115] According to an embodiment of the present disclosure, the above 7 time-frequency domain sub-features are combined into a long feature vector to obtain a complete time-frequency domain feature corresponding to the ISAL laser inverse synthetic aperture image.

[0116] According to an embodiment of the present disclosure, gradient transformation is performed based on scale-invariant features to obtain texture features, including: dividing an ISAL laser inverse synthetic aperture image into regions to obtain a plurality of local regions; for an i-th local region among the plurality of local regions, processing the scale-invariant features corresponding to each pixel point located in the i-th local region using a gradient algorithm to obtain a gradient magnitude and a gradient direction of each pixel point; determining an i-th gradient histogram corresponding to the i-th local region based on the gradient magnitude and the gradient direction of each pixel point; encoding the i-th local region according to the i-th gradient histogram to obtain an i-th local texture feature; and determining a texture feature according to the local texture features corresponding to each of the plurality of local regions.

[0117] According to an embodiment of the present disclosure, for each local area among multiple local areas, a gradient algorithm is used to process the scale-invariant features corresponding to each pixel point located in each local area to obtain the gradient amplitude and gradient direction of each pixel point, thereby determining the gradient histogram corresponding to each local area and obtaining the local texture features corresponding to each local area.

[0118] According to the embodiments of the present disclosure, the gradient amplitude represents the speed of the brightness or color change of the pixel in the ISAL laser inverse synthetic aperture image, and the gradient direction represents the direction of the brightness or color change of the pixel in the ISAL laser inverse synthetic aperture image. The gradient of the pixel can be represented by the gradient amplitude and the gradient direction.

[0119] According to an embodiment of the present disclosure, for the i-th local area, normalization and direction assignment are first performed to eliminate the influence of illumination and rotation; then the gradient algorithm is used to process the scale-invariant features corresponding to each pixel point located in the i-th local area to obtain the gradient amplitude and gradient direction between each pixel point and the adjacent pixel points, thereby determining the i-th gradient histogram corresponding to the i-th local area.

[0120] According to an embodiment of the present disclosure, local texture features characterize the changing properties of the structure and tissue arrangement in a local area, and the local texture features of multiple local areas are merged to determine the complete texture features corresponding to the ISAL laser inverse synthetic aperture image.

[0121] Figure 3 A flowchart of training a support vector machine model according to an embodiment of the present disclosure is shown.

[0122] like Figure 3 As shown, training the support vector machine model includes steps S310 to S330.

[0123] In step S310, a training sample is obtained.

[0124] In step S320, feature extraction is performed on the sample ISAL laser inverse synthetic aperture image to obtain sample time-frequency domain features and sample texture features.

[0125] In step S330, a support vector machine model is trained based on the sample time-frequency domain features, the sample texture features, and the classification labels to obtain a trained support vector machine model.

[0126] According to an embodiment of the present disclosure, the training samples include sample ISAL laser inverse synthetic aperture images and classification labels, and the classification labels represent quality assessment categories of the sample ISAL laser inverse synthetic aperture images, and the quality assessment categories include shooting posture categories and clarity categories.

[0127] According to an embodiment of the present invention, the shooting posture category and the clarity category are combined, and the classification labels are 24 subcategories, namely, front view - excellent, front view - good, front view - medium, front view - poor; side view - excellent, side view - good, side view - medium, side view - poor; top view - excellent, top view - good, top view - medium, top view - poor; bottom view - excellent, bottom view - good, bottom view - medium, bottom view - poor; rear view - excellent, rear view - good, rear view - medium, rear view - poor; strabismus - excellent, strabismus - good, strabismus - medium, strabismus - poor.

[0128] According to the embodiments of the present disclosure, a biorthogonal wavelet basis algorithm (Bior) is used to process a sample ISAL laser inverse synthetic aperture image to obtain sample time-frequency domain features; a feature transformation algorithm (RDSIFT) is used to process the position information of pixel points in the sample ISAL laser inverse synthetic aperture image to obtain sample directional features, and then obtain sample scale-invariant features; and a gradient transformation is performed on the sample scale-invariant features to obtain sample texture features.

[0129] According to an embodiment of the present disclosure, sample time-frequency domain features, sample texture features, and classification labels are input into a support vector machine model for training to obtain a trained support vector machine model.

[0130] According to an embodiment of the present disclosure, a one-to-one encoding method (One-vs-One, OvO) is used to convert a multi-category problem into multiple binary classification sub-problems, and multiple binary classifiers are trained. The sub-problems are independent of each other, and the binary classifier of each sub-problem is used to distinguish a sub-category. During prediction, the outputs of all sub-classifiers are combined to determine the final multi-category classification result.

[0131] According to an embodiment of the present disclosure, a support vector machine model is trained based on sample time-frequency domain features, sample texture features, and classification labels, and the trained support vector machine model includes: inputting sample time-frequency domain features and sample texture features into the support vector machine model, and outputting sample image quality assessment results; using a loss function to process the sample image quality assessment results and classification labels to obtain a loss value; and training the support vector machine model based on the loss value to obtain a trained support vector machine model.

[0132] According to an embodiment of the present disclosure, the sample image quality assessment result represents a prediction result of assessing the quality of a sample ISAL laser inverse synthetic aperture image.

[0133] According to an embodiment of the present disclosure, the loss function may be a binary classification loss function. According to the loss value, the parameters are adjusted to iteratively train the support vector machine model until the loss value satisfies a condition for stopping training.

[0134] Based on the ISAL image quality assessment method based on Bior and RDSIFT algorithms, the present disclosure also provides an ISAL image quality assessment device. Figure 4 The device is described in detail.

[0135] Figure 4 A structural block diagram of an ISAL image quality assessment device according to an embodiment of the present disclosure is shown.

[0136] like Figure 4As shown, the ISAL image quality assessment device 400 of this embodiment includes a first processing module 410 , a second processing module 420 , a convolution module 430 , a transformation module 440 and an assessment module 450 .

[0137] The first processing module 410 is used to process the ISAL laser inverse synthetic aperture image using a biorthogonal wavelet basis algorithm (Bior) to obtain time-frequency domain features. In one embodiment, the first processing module 410 can be used to perform the operation S110 described above, which will not be repeated here.

[0138] The second processing module 420 is used to process the position information of the pixel points in the ISAL laser inverse synthetic aperture image using a feature transformation algorithm (RDSIFT) to obtain directional features. In one embodiment, the second processing module 420 can be used to perform the operation S120 described above, which will not be repeated here.

[0139] The convolution module 430 is used to convolve the grayscale features and directional features of the ISAL laser inverse synthetic aperture image to obtain scale-invariant features. In one embodiment, the convolution module 430 can be used to perform the operation S130 described above, which will not be described in detail here.

[0140] The transformation module 440 is used to perform gradient transformation on the scale-invariant feature to obtain a texture feature. In one embodiment, the transformation module 440 can be used to perform the operation S140 described above, which will not be described in detail here.

[0141] The evaluation module 450 is used to fuse the time-frequency domain features and the texture features into the support vector machine model and output the image quality evaluation result. In one embodiment, the evaluation module 450 can be used to perform the operation S150 described above, which will not be repeated here.

[0142] According to an embodiment of the present disclosure, the second processing module 420 includes a first processing submodule, a second processing submodule and a third processing submodule.

[0143] The first processing submodule is used to process the position information of the pixel points in the ISAL laser inverse synthetic aperture image according to the Gaussian function to obtain the spatial features.

[0144] The second processing submodule is used to process the position information and the spatial features according to the radial basis function to obtain the radial basis features.

[0145] The third processing submodule is used to process the position information and the radial basis feature according to the directional filtering algorithm to obtain the directional feature.

[0146] According to an embodiment of the present disclosure, the first processing module 410 includes a fourth processing submodule, a fifth processing submodule, a sixth processing submodule, and a seventh processing submodule.

[0147] The fourth processing submodule is used to transform the ISAL laser inverse synthetic aperture image based on the biorthogonal wavelet basis algorithm (Bior) to obtain a first reference image, a first horizontal azimuth image, a first vertical azimuth image, and a first diagonal azimuth image, wherein the first reference image represents an image whose image grayscale value change frequency is less than a first threshold value, and the first horizontal azimuth image, the first vertical azimuth image, and the first diagonal azimuth image all represent images whose image grayscale value change frequency is greater than a second threshold value.

[0148] The fifth processing submodule is used to transform the first reference image based on the biorthogonal wavelet basis algorithm (Bior) to obtain a second reference image, a second horizontal azimuth image, a second vertical azimuth image, and a second diagonal azimuth image, wherein the second reference image represents an image whose image grayscale value change frequency is less than a first threshold, and the second horizontal azimuth image, the second vertical azimuth image, and the second diagonal azimuth image all represent images whose image grayscale value change frequency is greater than a second threshold.

[0149] The sixth processing submodule is used to perform quantization processing on the second reference image, the second horizontal azimuth image, the second vertical azimuth image, the second diagonal azimuth image, the first horizontal azimuth image, the first vertical azimuth image, and the first diagonal azimuth image, respectively, to obtain the second reference quantized image, the second horizontal azimuth quantized image, the second vertical azimuth quantized image, the second diagonal azimuth quantized image, the first horizontal azimuth quantized image, the first vertical azimuth quantized image, and the first diagonal azimuth quantized image.

[0150] The seventh processing submodule is used to extract features from the second reference quantized image, the second horizontal azimuth quantized image, the second vertical azimuth quantized image, the second diagonal azimuth quantized image, the first horizontal azimuth quantized image, the first vertical azimuth quantized image, and the first diagonal azimuth quantized image to obtain time-frequency domain features corresponding to the ISAL laser inverse synthetic aperture image.

[0151] According to an embodiment of the present disclosure, the sixth processing submodule includes a first processing unit, a second processing unit, a third processing unit, a fourth processing unit, a fifth processing unit and a sixth processing unit.

[0152] The first processing unit is used to process the pixel wavelet coefficients of the first diagonal orientation image according to the noise standard deviation algorithm to obtain a first noise standard deviation.

[0153] The second processing unit is used to process the first noise standard deviation and the number of pixels of the first diagonal orientation image according to a threshold algorithm to obtain a first threshold.

[0154] The third processing unit is used to perform quantization processing on the first horizontal azimuth image, the first vertical azimuth image, and the first diagonal azimuth image based on the first threshold value to obtain the first horizontal azimuth quantized image, the first vertical azimuth quantized image, and the first diagonal azimuth quantized image.

[0155] The fourth processing unit is used to process the pixel wavelet coefficients of the second diagonal orientation image according to the noise standard deviation algorithm to obtain a second noise standard deviation.

[0156] The fifth processing unit is used to process the second noise standard deviation and the number of pixels of the second diagonal orientation image according to a threshold algorithm to obtain a second threshold.

[0157] The sixth processing unit is used to quantize the pixel wavelet coefficients of the second reference image, the second horizontal azimuth image, the second vertical azimuth image, and the second diagonal azimuth image based on the second threshold value to obtain the second reference quantized image, the second horizontal azimuth quantized image, the second vertical azimuth quantized image, and the second diagonal azimuth quantized image.

[0158] According to an embodiment of the present disclosure, the seventh processing submodule includes a seventh processing unit and an eighth processing unit.

[0159] The seventh processing unit is used to extract time-frequency domain features of the second baseline quantized image, the second horizontal azimuth quantized image, the second vertical azimuth quantized image, the second diagonal azimuth quantized image, the first horizontal azimuth quantized image, the first vertical azimuth quantized image, and the first diagonal azimuth quantized image, respectively, to obtain multiple time-frequency domain sub-features, wherein the time-frequency domain sub-features include frequency band entropy information, frequency band energy information, frequency band average information, and frequency band standard deviation information.

[0160] The eighth processing unit is used to fuse multiple time-frequency domain sub-features to obtain the time-frequency domain features corresponding to the ISAL laser inverse synthetic aperture image.

[0161] According to the embodiments of the present invention, any one or more of the modules, submodules, units, and subunits, or at least part of the functions of any one of them can be implemented in one module. According to the embodiments of the present invention, any one or more of the modules, submodules, units, and subunits can be split into multiple modules for implementation. According to the embodiments of the present invention, any one or more of the modules, submodules, units, and subunits can be at least partially implemented as hardware circuits, such as field programmable gate arrays (FPGAs), programmable logic arrays (PLAs), systems on chips, systems on substrates, systems on packages, application specific integrated circuits (ASICs), or can be implemented by hardware or firmware in any other reasonable way of integrating or packaging circuits, or can be implemented in any one of the three implementation methods of software, hardware, and firmware, or in any appropriate combination of any of them. Alternatively, according to the embodiments of the present invention, one or more of the modules, submodules, units, and subunits can be at least partially implemented as computer program modules, and when the computer program modules are run, the corresponding functions can be performed.

[0162] For example, any of the first processing module 410, the second processing module 420, the convolution module 430, the transformation module 440, and the evaluation module 450 can be combined in one module / unit / sub-unit for implementation, or any of the modules / units / sub-units can be split into multiple modules / units / sub-units. Alternatively, at least part of the functions of one or more of these modules / units / sub-units can be combined with at least part of the functions of other modules / units / sub-units and implemented in one module / unit / sub-unit. According to an embodiment of the present disclosure, at least one of the first processing module 410, the second processing module 420, the convolution module 430, the transformation module 440, and the evaluation module 450 can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on a chip, a system on a substrate, a system on a package, an application specific integrated circuit (ASIC), or can be implemented by hardware or firmware such as any other reasonable way of integrating or packaging the circuit, or by any one of the three implementation methods of software, hardware, and firmware, or by a suitable combination of any of them. Alternatively, at least one of the first processing module 410, the second processing module 420, the convolution module 430, the transformation module 440 and the evaluation module 450 may be at least partially implemented as a computer program module, which may perform corresponding functions when executed.

[0163] Figure 5 A block diagram of an electronic device suitable for implementing the ISAL image quality assessment method based on the Bior and RDSIFT algorithms according to an embodiment of the present invention is shown.

[0164] Figure 5 The electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0165] like Figure 5 As shown, the computer electronic device 500 according to an embodiment of the present invention includes a processor 501, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 502 or the program loaded from the storage part 508 to the random access memory (RAM) 503. The processor 501 may include, for example, a general-purpose microprocessor (such as a CPU), an instruction set processor and / or a related chipset and / or a special-purpose microprocessor (for example, an application-specific integrated circuit (ASIC)), etc. The processor 501 may also include an on-board memory for caching purposes. The processor 501 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present invention.

[0166] In RAM 503, various programs and data required for the operation of electronic device 500 are stored. Processor 501, ROM 502 and RAM 503 are connected to each other via bus 504. Processor 501 performs various operations of the method flow according to the embodiment of the present invention by executing the program in ROM 502 and / or RAM 503. It should be noted that the program can also be stored in one or more memories other than ROM 502 and RAM 503. Processor 501 can also perform various operations of the method flow according to the embodiment of the present invention by executing the program stored in one or more memories.

[0167] Optionally, the electronic device 500 may further include an input / output (I / O) interface 505, which is also connected to the bus 504. The electronic device 500 may further include one or more of the following components connected to the input / output (I / O) interface 505: an input portion 506 including a keyboard, a mouse, etc.; an output portion 507 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage portion 508 including a hard disk, etc.; and a communication portion 509 including a network interface card such as a LAN card, a modem, etc. The communication portion 509 performs communication processing via a network such as the Internet. A drive 510 is also connected to the input / output (I / O) interface 505 as needed. A removable medium 511, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 510 as needed, so that a computer program read therefrom is installed into the storage portion 508 as needed.

[0168] Optionally, the method flow according to the embodiment of the present invention can be implemented as a computer software program. For example, an embodiment of the present invention includes a computer program product, which includes a computer program carried on a computer-readable storage medium, and the computer program contains a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 509, and / or installed from the removable medium 511. When the computer program is executed by the processor 501, the above-mentioned functions defined in the system of the embodiment of the present invention are executed. Optionally, the system, equipment, device, module, unit, etc. described above can be implemented by a computer program module.

[0169] The present invention also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiment; or may exist independently without being assembled into the device / apparatus / system. The above computer-readable storage medium carries one or more programs, and when the above one or more programs are executed, the ISAL image quality assessment method based on the Bior and RDSIFT algorithms according to the embodiment of the present invention is implemented.

[0170] Optionally, the computer-readable storage medium may be a non-volatile computer-readable storage medium. For example, it may include, but is not limited to: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, apparatus, or device.

[0171] For example, optionally, the computer-readable storage medium may include the ROM 502 and / or the RAM 503 described above and / or one or more memories other than the ROM 502 and the RAM 503 .

[0172] An embodiment of the present invention also includes a computer program product, which includes a computer program, which contains program code for executing the method provided by the embodiment of the present invention. When the computer program product runs on an electronic device, the program code is used to enable the electronic device to implement the ISAL image quality assessment method based on the Bior and RDSIFT algorithms provided by the embodiment of the present invention.

[0173] When the computer program is executed by the processor 501, the above functions defined in the system / device of the embodiment of the present invention are performed. Optionally, the system, device, module, unit, etc. described above can be implemented by a computer program module.

[0174] In one embodiment, the computer program may rely on tangible storage media such as optical storage devices, magnetic storage devices, etc. In another embodiment, the computer program may also be transmitted and distributed in the form of signals on a network medium, and downloaded and installed through the communication part 509, and / or installed from the removable medium 511. The program code contained in the computer program may be transmitted using any appropriate network medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.

[0175] Optionally, the program code for executing the computer program provided by the embodiment of the present invention can be written in any combination of one or more programming languages, specifically, these computing programs can be implemented using high-level process and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, such as Java, C++, python, "C" language or similar programming languages. The program code can be executed entirely on the user computing device, partially on the user device, partially on the remote computing device, or entirely on the remote computing device or server. In the case of a remote computing device, the remote computing device can be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computing device (e.g., using an Internet service provider to connect through the Internet).

[0176] The flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each box in the flowchart or block diagram may represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of the code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box may also occur in an order different from that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flowchart, and the combination of boxes in the block diagram or flowchart, can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions. It can be understood by those skilled in the art that the features recorded in the various embodiments of the present disclosure can be combined and / or combined in a variety of ways, even if such a combination or combination is not explicitly recorded in the present disclosure. In particular, without departing from the spirit and teaching of the present disclosure, the features described in the various embodiments of the present disclosure may be combined and / or combined in a variety of ways. All of these combinations and / or combinations fall within the scope of the present disclosure.

[0177] The embodiments of the present disclosure are described above. However, these embodiments are only for illustrative purposes and are not intended to limit the scope of the present disclosure. Although the embodiments are described above, this does not mean that the measures in the various embodiments cannot be used in combination to advantage. Without departing from the scope of the present disclosure, those skilled in the art may make a variety of substitutions and modifications, which should all fall within the scope of the present disclosure.

Claims

1. An ISAL image quality assessment method based on Bior and RDSIFT algorithms, characterized in that: The method comprises: The ISAL laser inverse synthetic aperture image is processed using the biorthogonal wavelet basis algorithm (Bior) to obtain the time-frequency domain features. Using a feature transformation algorithm (RDSIFT) to process the position information of the pixel points in the ISAL laser inverse synthetic aperture image to obtain a directional feature; Convolving the grayscale feature of the ISAL laser inverse synthetic aperture image and the directional feature to obtain a scale-invariant feature; Performing gradient transformation on the scale-invariant feature to obtain a texture feature; The time-frequency domain features and the texture features are fused and input into a support vector machine model to output an image quality assessment result.

2. The method according to claim 1, characterized in that The ISAL laser inverse synthetic aperture image is processed using a feature transformation algorithm (RDSIFT) to obtain directional features including: Processing the position information of the pixel points in the ISAL laser inverse synthetic aperture image according to the Gaussian function to obtain the spatial features; Processing the position information and the spatial features according to a radial basis function to obtain a radial basis feature; The position information and the radial basis feature are processed according to a directional filtering algorithm to obtain the directional feature.

3. The method according to claim 1, characterized in that: The ISAL laser inverse synthetic aperture image is processed using the biorthogonal wavelet basis algorithm (Bior), and the time-frequency domain features obtained include: The ISAL laser inverse synthetic aperture image is transformed based on a biorthogonal wavelet basis algorithm (Bior) to obtain a first reference image, a first horizontal azimuth image, a first vertical azimuth image, and a first diagonal azimuth image, wherein the first reference image represents an image whose image grayscale value change frequency is less than a first threshold, and the first horizontal azimuth image, the first vertical azimuth image, and the first diagonal azimuth image all represent images whose image grayscale value change frequency is greater than a second threshold; Transforming the first reference image based on the biorthogonal wavelet basis algorithm (Bior) to obtain a second reference image, a second horizontal azimuth image, a second vertical azimuth image, and a second diagonal azimuth image, wherein the second reference image represents an image whose grayscale value change frequency is less than a first threshold, and the second horizontal azimuth image, the second vertical azimuth image, and the second diagonal azimuth image all represent images whose grayscale value change frequency is greater than a second threshold; Respectively perform quantization processing on the second reference image, the second horizontal azimuth image, the second vertical azimuth image, the second diagonal azimuth image, the first horizontal azimuth image, the first vertical azimuth image, and the first diagonal azimuth image to obtain a second reference quantized image, a second horizontal azimuth quantized image, a second vertical azimuth quantized image, a second diagonal azimuth quantized image, a first horizontal azimuth quantized image, a first vertical azimuth quantized image, and a first diagonal azimuth quantized image; Feature extraction is performed on the second reference quantized image, the second horizontal azimuth quantized image, the second vertical azimuth quantized image, the second diagonal azimuth quantized image, the first horizontal azimuth quantized image, the first vertical azimuth quantized image, and the first diagonal azimuth quantized image to obtain time-frequency domain features corresponding to the ISAL laser inverse synthetic aperture image.

4. The method according to claim 3, characterized in that The second reference image, the second horizontal azimuth image, the second vertical azimuth image, the second diagonal azimuth image, the first horizontal azimuth image, the first vertical azimuth image, and the first diagonal azimuth image are quantized respectively to obtain a second reference quantized image, a second horizontal azimuth quantized image, a second vertical azimuth quantized image, a second diagonal azimuth quantized image, a first horizontal azimuth quantized image, a first vertical azimuth quantized image, and a first diagonal azimuth quantized image, including: Processing the pixel wavelet coefficients of the first diagonal orientation image according to a noise standard deviation algorithm to obtain a first noise standard deviation; Processing the first noise standard deviation and the number of pixels of the first diagonal orientation image according to a threshold algorithm to obtain a first threshold; Based on the first threshold, quantize the first horizontal azimuth image, the first vertical azimuth image, and the first diagonal azimuth image to obtain the first horizontal azimuth quantized image, the first vertical azimuth quantized image, and the first diagonal azimuth quantized image; Processing the pixel wavelet coefficients of the second diagonal orientation image according to the noise standard deviation algorithm to obtain a second noise standard deviation; Processing the second noise standard deviation and the number of pixels of the second diagonal orientation image according to the threshold algorithm to obtain a second threshold; Based on the second threshold, the pixel wavelet coefficients of the second reference image, the second horizontal azimuth image, the second vertical azimuth image, and the second diagonal azimuth image are quantized to obtain the second reference quantized image, the second horizontal azimuth quantized image, the second vertical azimuth quantized image, and the second diagonal azimuth quantized image.

5. The method according to claim 3, characterized in that: Performing feature extraction on the second reference quantized image, the second horizontal azimuth quantized image, the second vertical azimuth quantized image, the second diagonal azimuth quantized image, the first horizontal azimuth quantized image, the first vertical azimuth quantized image, and the first diagonal azimuth quantized image to obtain time-frequency domain features corresponding to the ISAL laser inverse synthetic aperture image includes: Performing time-frequency domain feature extraction on the second reference quantized image, the second horizontal azimuth quantized image, the second vertical azimuth quantized image, the second diagonal azimuth quantized image, the first horizontal azimuth quantized image, the first vertical azimuth quantized image, and the first diagonal azimuth quantized image respectively to obtain a plurality of time-frequency domain sub-features, wherein the time-frequency domain sub-features include frequency band entropy information, frequency band energy information, frequency band average information, and frequency band standard deviation information; A plurality of the time-frequency domain sub-features are fused to obtain the time-frequency domain features corresponding to the ISAL laser inverse synthetic aperture image.

6. The method according to claim 1, characterized in that Based on the scale-invariant feature, gradient transformation is performed to obtain texture features including: Performing region division on the ISAL laser inverse synthetic aperture image to obtain a plurality of local regions; For an i-th local area among the multiple local areas, use a gradient algorithm to process the scale-invariant feature corresponding to each pixel point located in the i-th local area to obtain a gradient magnitude and a gradient direction of each pixel point; Determining an i-th gradient histogram corresponding to the i-th local area based on the gradient magnitude and gradient direction of each pixel point; Encoding the i-th local area according to the i-th gradient histogram to obtain an i-th local texture feature; The texture feature is determined according to local texture features corresponding to each of the plurality of local areas.

7. The method according to claim 1, characterized in that The support vector machine model is trained based on the following operations: Acquire a training sample, wherein the training sample includes a sample ISAL laser inverse synthetic aperture image and a classification label, the classification label represents a quality assessment category of the sample ISAL laser inverse synthetic aperture image, and the quality assessment category includes a shooting posture category and a clarity category; Extracting features of the sample ISAL laser inverse synthetic aperture image to obtain the sample time-frequency domain features and the sample texture features; Based on the sample time-frequency domain features, the sample texture features, and the classification labels, a support vector machine model is trained to obtain a trained support vector machine model.

8. The method according to claim 7, characterized in that Based on the sample time-frequency domain features, the sample texture features, and the classification labels, a support vector machine model is trained to obtain a trained support vector machine model including: Based on the sample time-frequency domain features and the sample texture features, the sample is input into the support vector machine model, and a sample image quality assessment result is output; Processing the sample image quality assessment result and the classification label using a loss function to obtain a loss value; According to the loss value, a support vector machine model is trained to obtain a trained support vector machine model.

9. An ISAL image quality assessment device, characterized in that: The device comprises: The first processing module is used to process the ISAL laser inverse synthetic aperture image using a biorthogonal wavelet basis algorithm (Bior) to obtain time-frequency domain features; The second processing module is used to process the position information of the pixel points in the ISAL laser inverse synthetic aperture image by using a feature transformation algorithm (RDSIFT) to obtain a directional feature; A convolution module, used for convolving the grayscale feature of the ISAL laser inverse synthetic aperture image and the directional feature to obtain a scale-invariant feature; A transformation module, used for performing gradient transformation on the scale-invariant feature to obtain a texture feature; The evaluation module is used to fuse the time-frequency domain features and the texture features into a support vector machine model and output an image quality evaluation result.

10. An electronic device comprising: one or more processors; a memory for storing one or more programs, When one or more programs are executed by one or more processors, the one or more processors implement the method of any one of claims 1 to 8.

Images