UTOD-based underwater imaging quality evaluation method

Abstract

The application discloses a kind of underwater imaging quality evaluation methods based on UTOD, belong to underwater detection and imaging technical field, this method includes obtaining triangle target feature, and the image obtained is processed by human eye visual system HVS simulation module;Underwater triangle direction discrimination threshold UTOD model is constructed, and UTOD model is used to distinguish the direction of triangle target;Contrast-sharpness comprehensive index UICS is constructed, including underwater sharpness index and underwater contrast index;Based on four options forced choice 4AFC test, the direction discrimination correct probability corresponding to different UICS values is counted, and the relationship curve of UICS and correct probability is fitted by psychometric function Weibull;Triangle target of different size is repeatedly tested, and UTOD curve is generated.The application realizes automatic target detection and direction identification, and improves the accuracy, stability and objectivity of evaluation index.

Description

Technical Field

[0001] This invention relates to the field of underwater detection and imaging technology, specifically to an underwater imaging quality evaluation method based on UTOD. Background Technology

[0002] With the development of underwater detection and imaging technologies, how to scientifically and effectively evaluate underwater image quality has become a key issue. Existing underwater image quality evaluation methods can be broadly divided into two categories: one is image-based evaluation methods, which quantify quality by analyzing the features of the image content itself; the other is performance model-based evaluation methods, which use visual system models or task-oriented models to assess the image's performance in tasks such as detection and recognition. However, even though performance model-based methods outperform quality evaluation methods that solely rely on image features in terms of systematicity, stability, and task relevance, there is currently a lack of dedicated performance evaluation models for underwater imaging systems. Most research remains at the level of image-based indicators, which is insufficient to meet the evaluation needs of complex underwater tasks and easily leads to inconsistencies between subjective and objective assessments.

[0003] Currently, the Triangle Orientation Discrimination Threshold (TOD) method, widely used in traditional optical systems, is a measurement method based on triangular targets. It quantifies the human eye's ability to recognize the target's orientation for image recognition. While this method effectively measures image quality and is easy to implement, in underwater imaging scenarios, the traditional TOD model relies on intensity contrast (…). As a core parameter, it is difficult to achieve reasonable quantification of underwater target information under different turbidity levels. In addition, traditional TOD models rely on subjective human visual identification in target orientation discrimination tasks. Due to the large workload, time-consuming and tedious process, observers are prone to fatigue, leading to instability in subjective psychology and objective state, making the test results unable to accurately reflect the actual situation. Summary of the Invention

[0004] The purpose of this invention is to propose an underwater imaging quality evaluation method based on UTOD. By fusing HVS characteristics and corner-edge joint features, a contrast-sharpness integrated index (UICS) is constructed, which realizes automated target detection and orientation recognition, and improves the consistency between subjective and objective aspects.

[0005] To achieve the above objectives, this invention proposes an underwater imaging quality assessment method based on UTOD, the steps of which are as follows:

[0006] Step S1: Collect the feature information of the triangular target and process the acquired image through the human visual system HVS simulation module;

[0007] Step S2, constructing an underwater triangle direction discrimination threshold UTOD model, and using the UTOD model to discriminate the direction of the triangle target;

[0008] Step S3, constructing a contrast-sharpness comprehensive index UICS, including an underwater sharpness index and an underwater contrast index;

[0009] Step S4, calculating the contrast-sharpness comprehensive index;

[0010] Step S5, using a four-alternative forced choice 4AFC method to discriminate the direction of the image, after counting the direction discrimination correct probability corresponding to different UICS values, fitting the relationship curve between UICS and the correct probability through a psychometric function Weibull;

[0011] Step S6, repeatedly testing a plurality of triangle targets of different sizes to generate a UTOD curve.

[0012] Preferably, in step S1, the specific steps are as follows:

[0013] Step S11, using a human visual system HVS simulation module to process the image, and the specific steps are as follows:

[0014] Step S111, simulating visual nonlinear effects, based on the nonlinear luminance response conversion of the Weber-Fechner law, the calculation formula is as follows:

[0015] ;

[0016] Wherein, is the perceived brightness, is a constant related to the average brightness of the entire image, is the objective brightness, is a basic response threshold representing the visual system;

[0017] Step S112, simulating the multi-channel structure, using a band-pass filter with specific spatial frequency and direction characteristics to simulate the modulation effect of the multi-channel signal, and the specific steps are as follows:

[0018] Using a Gabor filter bank to extract directional texture features, the specific calculation formula is as follows:

[0019] ;

[0020] ;

[0021] ;

[0022] is a Gabor filter with a direction of ; is the coordinate of the image pixel point, is the coordinate of the image pixel point after coordinate transformation, is the filter direction, is the center frequency, is the standard deviation;

[0023] Four-channel filtering is performed on the input image, and the specific calculation formula is as follows:

[0024] ;

[0025] wherein, is the convolution result of the input underwater image and the Gabor filter with the direction being ;

[0026] The key information of the image in each direction is extracted, and the energy of each direction is extracted. The specific calculation formula is as follows:

[0027] ;

[0028] wherein, is the energy value in the direction, is the total number of image pixels;

[0029] Step S113, compare the sensitivity function CSF and fuse the direction sub-band, and the specific steps are as follows:

[0030] The contrast sensitivity equation CSF is used to describe the correlation between the human visual perception of image contrast and background brightness and spatial frequency. The specific calculation formula is as follows:

[0031] ;

[0032] ;

[0033] wherein, is the contrast sensitivity function value, is the frequency component in the horizontal direction of the image, is the frequency component in the vertical direction, is the spatial frequency;

[0034] The specific calculation formula for weighting each direction sub-band is as follows:

[0035] ;

[0036] wherein, is the energy value in the direction after CSF weighting;

[0037] The HVS energy feature image is obtained by fusing each direction feature map, and the specific calculation formula is as follows:

[0038] ;

[0039] wherein, is the HVS energy feature image;

[0040] Step S12, the target detection method based on the joint feature extraction of the corner and the edge obtains the triangular target feature, and the improved strategy of the Harris corner detection is utilized, and the specific steps are as follows:

[0041] Step S121, detection is performed according to the characteristics of the triangular vertex, and the specific calculation formula is as follows:

[0042] ;

[0043] ;

[0044] wherein, is the determinant of the matrix , is the trace of the matrix , is an empirical constant, is the Hessian matrix, is the gradient of in the direction, is the gradient of in the direction, is the Harris corner response function; Step S122, the obtained numerous corner points are screened, and the specific screening rule is as follows:

[0045] , and the first three maximum response points are reserved;

[0046] , and the first three maximum response points are reserved;

[0047] wherein, is a set dynamic threshold, and the first three maximum response points are forcibly reserved;

[0048] Step S13, the target detection method based on the joint feature extraction of the corner and the edge obtains the triangular target feature, and the Canny edge detection is optimized, and the specific steps are as follows:

[0049] Step S131, according to the actual underwater imaging environment, a double-threshold rule is set, and the specific calculation formula is as follows:

[0050] ;

[0051] wherein, is the maximum gradient in the image, is a low threshold, is a high threshold;​​

[0052] Step S132, based on the characteristic that real corner points are usually located on or close to the edge, the spatial relationship between edge points and corner points is used to eliminate false corner points, for each Harris corner point, the minimum Euclidean distance between it and all Canny edge points is calculated:

[0053] ;

[0054] wherein, is the minimum Euclidean distance between each Harris corner point and all Canny edge points, is the coordinate of the corner point, is the coordinate of the edge point, represents the minimum value in all edge point coordinates .

[0055] Preferably, in step S2, the specific steps are as follows:

[0056] Step S21, determine the pointing direction of the triangular target, calculate the geometric center based on the position information of the three vertices, and the specific calculation formula is:

[0057] ;

[0058] wherein, is the geometric center of the triangular target, is the horizontal coordinate of the triangular vertex, is the vertical coordinate of the triangular vertex, is the number of triangular vertices;

[0059] Step S22, taking the center as the reference, divide the triangular pattern into two parts in the horizontal and vertical directions, and the corresponding areas are , and , respectively, calculate the area ratio of each part to obtain the area factor , , and the specific calculation formula is as follows:

[0060] ;

[0061] wherein, is the area factor of the upper and lower parts, is the area factor of the left and right parts, is the area of the upper half of the triangular pattern, is the area of the lower half of the triangular pattern, is the area of the left half of the triangular pattern, This represents the area of ​​the right half of the triangle.

[0062] Step S23: Determine the direction of the triangular target according to the determination rules. The specific determination rules are as follows:

[0063] ;

[0064] in, This is the tolerance value.

[0065] Preferably, in step S3, the contrast-clarity integrated index (UICS) is constructed, and the specific steps are as follows:

[0066] Step S31: Calculate the underwater clarity index. The specific calculation steps are as follows:

[0067] Step S311: Apply to the image The operator calculates the gradients in the horizontal and vertical directions to obtain the edge intensity map, as shown in the following formula:

[0068] ;

[0069] in, The gradient in the horizontal direction, In the horizontal direction Operator, The original target image. The gradient in the vertical direction, In the vertical direction Operator;

[0070] Step S312: Define the calculated edge intensity map as the gradient magnitude, as shown in the following formula:

[0071] ;

[0072] in, for Edge intensity image;

[0073] Step S313: Multiply the edge intensity map pixel by pixel with the original image to retain the intensity information of the edge region. The formula is as follows:

[0074] ;

[0075] in, This is a grayscale enhanced image after edge intensity weighting;

[0076] Step S314: Calculate relative contrast based on block division to obtain The formula is as follows:

[0077] ;

[0078] ;

[0079] wherein, is the underwater sharpness index, is the edge mapping energy, is the number of blocks in the vertical direction, is the number of blocks in the horizontal direction, is the vertical block index, is the horizontal block index, is the maximum gray value in the i-th row, column block, is the minimum gray value in the i-th row, column block;

[0080] Step S32, calculating the underwater contrast index , the formula is as follows:

[0081] ;

[0082] ;

[0083] wherein, is the underwater contrast index, is the average Michelson contrast in the local region of the image, is the multiplicative aggregation operation under PLIP, is the difference operation under PLIP, is the additive aggregation operation under PLIP.

[0084] Preferably, in step S4, the specific calculation formula is as follows:

[0085] ;

[0086] wherein, is the contrast-sharpness comprehensive index, and are weight factors.

[0087] Preferably, in step S5, the four-alternative forced choice (4AFC) method is used for image direction discrimination, the correct probability of direction discrimination corresponding to different UICS values is counted, and the calculation formula is as follows:

[0088] ;

[0089] wherein, is the correct probability of direction discrimination under different UICS, is the correct judgment deviation value, is the probability of random guessing when the direction is difficult to distinguish, is the UICS index threshold, is the slope of the curve.

[0090] Preferably, in step S6, repeated tests are performed on multiple triangular targets of different sizes, and finally a UTOD curve is generated, specifically: using a nonlinear least squares method to fit the experimental data, determining the parameters and the optimal solution of , extracting the UICS index threshold , and finally generating a UTOD curve with triangular angular spatial frequency as the horizontal coordinate, and the vertical coordinate.

[0091] Therefore, the present application proposes a UTOD-based underwater imaging quality evaluation method, which has the following beneficial effects:

[0092] (1) The task-oriented mechanism and visual perception characteristics are fused to achieve consistent subjective and objective evaluation results, and the image quality evaluation is associated with recognition, detection and other task performances, thereby effectively solving the problem that the evaluation index is disconnected with the actual use effect in traditional methods.

[0093] (2) A multi-dimensional information quantity perception model is constructed, an information quantity evaluation model is established, and the effective information useful for the task in the underwater image is quantitatively described, thereby enhancing the reliability and interpretability of the evaluation results.

[0094] (3) The index is designed for the target detection task, the efficient and objective automatic discrimination of the target direction is realized, the target feature extraction mechanism is constructed in combination with the direction-sensitive characteristics, and the rapid extraction and automatic discrimination of the target direction information in the complex underwater scene are supported.

[0095] The technical solutions of the present application will be further described in detail below with the aid of the drawings and embodiments.

[0096] Figure 1 is a step method flowchart of the UTOD-based underwater imaging quality evaluation method of the present application;

[0097] Figure 2 is a specific implementation case block diagram of the UTOD-based underwater imaging quality evaluation method of the present application;

[0098] Figure 3 is a schematic diagram for calculating the area factor of a triangular target; wherein Figure 3 (a) in the above is horizontally divided into upper and lower parts, Figure 3 (b) in the above is vertically divided into left and right parts;

[0099] Figure 4Fig. 1 is a schematic diagram of a triangular spline for four possible directions (up, down, left, right);

[0100] Figure 5 Fig. 2 is a schematic diagram of a direction recognition result of a triangular target, wherein, Figure 5 Fig. 2(a) is an example of an input image, Figure 5 Fig. 2(b) is a direction recognition result;

[0101] Figure 6 Fig. 3 is a schematic diagram of a Weibull function fitting curve and a UTOD curve, wherein, Figure 6 Fig. 3(a) is a Weibull function fitting curve, Figure 6 Fig. 3(b) is a UTOD curve;

[0102] Figure 7 Fig. 4 is a schematic diagram of target imaging effects under different algorithms, wherein, Figure 7 Fig. 4(a) is an original image of target imaging, Figure 7 Fig. 4(b) is a target imaging effect under a DCP algorithm, Figure 7 Fig. 4(c) is a target imaging effect under a GPLPF algorithm, Figure 7 Fig. 4(d) is a target imaging effect under a PDiff algorithm, Figure 7 Fig. 4(e) is a target imaging effect under a Chen algorithm, Figure 7 Fig. 4(f) is a target imaging effect under a PD2D algorithm;

[0103] Figure 8 Fig. 5 is a schematic diagram of UTOD curves of imaging results under different algorithms. DETAILED DESCRIPTION

[0104] The technical solutions of the present application are further described below by means of the accompanying drawings and examples.

[0105] Unless otherwise defined, the technical terms or scientific terms used in the present application shall have the usual meanings understood by those skilled in the art to which the present application belongs.

[0106] Example 1

[0107] As shown in Figures 1-2 , the present application provides a UTOD-based underwater imaging quality evaluation method, and the steps are as follows:

[0108] S1, obtain triangular target features, and process the image using a human visual system (HVS) simulation module, and the specific steps are as follows:

[0109] S11, process the image using a human visual system (HVS) simulation module, and the specific steps are as follows:

[0110] S111: Simulate visual nonlinear effects, nonlinear luminance response conversion based on Weber-Fechner law, the calculation formula is as follows:

[0111] ;

[0112] Wherein, is the perceived brightness, that is, the actual brightness level experienced by the human eye after the nervous system processing; is the objective brightness, which is the inherent brightness value of the actual scene or image obtained by high-precision optical measuring instrument; is the basic response threshold representing the visual system, reflecting the stable level of visual perception without external stimulus change, which is set to in the experiment; is a constant related to the average brightness of the whole image, whose value will be dynamically adjusted with the change of the overall brightness environment of the image, thereby affecting the sensitivity of the human eye to brightness change;

[0113] S112: Simulate multi-channel structure, simulate the modulation effect of multi-channel signals by using band-pass filter with specific spatial frequency and direction characteristics, the specific steps are as follows:

[0114] Gabor filter set is used to extract directional texture features, the specific calculation formula is as follows:

[0115] ;

[0116] ;

[0117] ;

[0118] is the image pixel coordinate, is the image pixel coordinate after coordinate transformation, is the filter direction, respectively 0°, 45°, 90°, 135°, is the center frequency, taking , corresponding to the medium frequency sensitive interval, is the standard deviation, taking , used to control the bandwidth, its main function is to suppress high-frequency noise;

[0119] Four-channel filtering is performed on the input image, and the specific calculation formula is as follows:

[0120] ;

[0121] Wherein, is the input underwater image, is the Gabor filter with direction , is the convolution result of the input underwater image and the Gabor filter with the direction of ;

[0122] The key information of the image in each direction is extracted, and the energy of each direction is extracted. The specific calculation formula is:

[0123] ;

[0124] wherein, is the energy value in the direction of , and is the total number of image pixels, which is the normalization molecule, is the convolution result of the input underwater image and the Gabor filter with the direction of , by squaring , the phase information can be effectively eliminated, and only the energy feature is retained, which simulates the nonlinear integration of complex cells to the direction energy, is the image pixel point coordinate;

[0125] S113: Contrast sensitivity function (CSF) weighting and fusing direction subbands, the specific steps are as follows:

[0126] The contrast sensitivity equation (CSF) is used to describe the correlation between the perception of human eye vision to image contrast and background brightness and spatial frequency. The specific calculation formula is:

[0127] ;

[0128] ;

[0129] wherein, is the contrast sensitivity function value, is the spatial frequency, which is cycles / degree, and it is calculated from the frequency spectrum information after Fourier transform, is the frequency component in the horizontal direction of the image, is the frequency component in the vertical direction. The spatial CSF has a bandpass characteristic in the spatial frequency, which describes the HVS sensitivity in each frequency domain component;

[0130] The specific calculation formula for weighting each direction subband is:

[0131] ;

[0132] wherein, is the energy value in the direction of weighted by CSF, is the contrast sensitivity function value, is the spatial frequency;

[0133] By using the CSF function for weighting, we can simulate the differences in the sensitivity of the human eye to contrast at different spatial frequencies and directions, so that the processing results of each directional sub-band conform to the rules of human subjective perception.

[0134] The HVS energy feature image is obtained by fusing feature maps from different directions. The specific calculation formula is as follows:

[0135] ;

[0136] in, HVS energy feature image;

[0137] S12: Extract features from the triangular target. A target detection method based on corner-edge joint feature extraction is used, employing an improved Harris corner detection strategy. The specific steps are as follows:

[0138] S121: An improved strategy using Harris corner detection is employed, which performs detection based on the characteristics of triangle vertices. The specific calculation formula is as follows:

[0139] ;

[0140] ;

[0141] in, For matrix The determinant, For matrix traces, This is an empirical constant, typically ranging from 0.04 to 0.06; here it is taken as [value missing]. , for exist Gradient of direction, for exist gradient of direction, This refers to the Harris corner response function;

[0142] according to Value determines the type of points in an image: If If the value is large and positive, the point is likely a corner point; if If the value is small and negative, the point may be an edge point; if If the value approaches zero, then the point is in a flat region;

[0143] S122: After obtaining numerous corner points, a selection process is performed to determine the corner points most likely to be triangle vertices. The specific selection rules are as follows:

[0144] , and the first three maximum response points are reserved;

[0145] is a set dynamic threshold, the dynamic threshold is set as Most of the noise false corners can be effectively filtered out, while the first three maximum response points are forced to be reserved, realizing the constraint of the number of vertices and preventing the number of response points from being less than three;

[0146] S13: Extract the features of the triangular target, based on the corner-edge joint feature extraction target detection method, optimize the Canny edge detection, the specific steps are as follows:

[0147] S131: According to the actual underwater imaging environment, set the double threshold rule, the specific calculation formula is:

[0148] ;

[0149] Wherein, is the maximum gradient value in the image, is the low threshold, is the high threshold;

[0150] The low threshold is used to retain some weak but possible real edge parts, such as blurred boundaries in turbid water, to prevent the triangular contour edge from breaking; the high threshold is to ensure that the obvious and strong edge can be continuously presented, while suppressing those isolated noise points generated by noise;

[0151] If the gradient amplitude , the point is determined as an edge point and directly reserved; if the gradient amplitude , the point is excluded as a non-edge point, realizing noise suppression; if the gradient amplitude , it is only reserved when the point is connected to a strong edge, that is, hysteresis threshold processing;

[0152] S132: Based on the characteristics that real corners are usually located on or near the edge, the spatial relationship between edge points and corners is used to eliminate false corners. For each Harris corner, calculate the minimum Euclidean distance between it and all Canny edge points:

[0153] ;

[0154] Wherein, is the minimum Euclidean distance between each Harris corner and all Canny edge points, is the corner coordinate, is the edge point coordinate, indicates that all edge point coordinates Take the minimum value;

[0155] If greater than 5 pixels, it is determined that the Harris corner point is a false corner point caused by noise, which is excluded, and finally 3 corner points are reserved.

[0156] S2, construct an underwater triangle direction discrimination threshold UTOD model to discriminate the triangle target direction, the steps are as follows:

[0157] S21: calculate the geometric center based on the position information of the triangle target vertex, the specific calculation formula is as follows:

[0158] ;

[0159] Wherein, is the geometric center of the triangle target, is the horizontal coordinate of the triangle vertex, is the vertical coordinate of the triangle vertex, is the number of triangle vertices;

[0160] S22: taking the center of gravity as the reference, the triangle figure is divided into upper and lower parts and left and right parts (such as Figure 3 ) according to the horizontal and vertical directions, and the corresponding areas are respectively recorded as , and , , the area ratio of each part is calculated to obtain the area factor , , which is used as a parameter to describe the direction characteristics of the triangle, and the specific calculation formula is as follows:

[0161] ;

[0162] Wherein is the area factor of the upper and lower parts, is the area factor of the left and right parts, is the area of the upper half of the triangle figure, is the area of the lower half of the triangle figure, is the area of the left half of the triangle figure, is the area of the right half of the triangle figure;

[0163] In the UTOD model measurement system, a uniform target with an equilateral triangle as the center is used as a key measurement element, and the triangle has four standard directions, namely up, down, left and right, and the direction difference is realized by the vertex pointing, and the specific direction is shown in Figure 4 ; taking the triangle target pointing upward as an example, in the ideal state, , In actual application scenarios, to enhance the fault tolerance of direction recognition, a tolerance value is introduced for adjustment, that is, when it can be judged as upward direction, and the tolerance value in the application is set to It is determined by collecting equilateral triangle samples in different scenarios and measuring the actual distribution of and ;

[0164] S23: Determine the direction of the triangle target according to the determination rule, and the specific determination rule is as follows:

[0165] ;

[0166] Wherein, is the tolerance value, ; without relying on reference images, the actual direction can be distinguished by the geometric characteristics of the triangle itself, and the result is shown in Figure 5 .

[0167] S3, construct a contrast-sharpness comprehensive index UICS, including underwater sharpness index and underwater contrast index, and the steps are as follows:

[0168] S31: Calculate the underwater sharpness index , and the specific calculation steps are as follows:

[0169] S311: Apply operator to the image to calculate the gradient in the horizontal and vertical directions to obtain the edge intensity map, and the specific calculation formula is as follows:

[0170] ;

[0171] Wherein, is the gradient in the horizontal direction, is the operator in the horizontal direction, is the original target image, is the gradient in the vertical direction, is the operator in the vertical direction;

[0172] S312: Define the calculated edge intensity map as the gradient amplitude, and the specific calculation formula is as follows:

[0173] ;

[0174] Wherein, is the edge intensity image of ;

[0175] S313: multiply the edge intensity map with the original image pixel by pixel to keep the intensity information of the edge region, and the specific calculation formula is as follows:

[0176] ;

[0177] wherein, is the gray scale enhanced image weighted by the edge intensity;

[0178] S314: calculate the relative contrast based on the block again to obtain , and the specific calculation formula is as follows:

[0179] ;

[0180] ;

[0181] wherein, is the underwater clarity index, is the edge mapping energy, is the number of blocks in the vertical direction, is the number of blocks in the horizontal direction, is the vertical block index, is the horizontal block index, is the maximum gray scale value in the row, column block, is the minimum gray scale value in the row, column block;

[0182] S32: calculate the underwater contrast index , and the specific calculation formula is as follows:

[0183] ;

[0184] ;

[0185] wherein, is the underwater contrast index, is the average Michelson contrast in the local region of the image, is the multiplicative aggregation operation under PLIP, is the difference operation under PLIP, is the additive aggregation operation under PLIP;

[0186] The image is divided into blocks, and the entropy operation is introduced into the traditional Agaian enhancement metric AMEE, and the PLIP operation provides a nonlinear representation consistent with human visual perception.

[0187] S4, calculate the contrast-clearness comprehensive index, the specific calculation formula is as follows:

[0188] ;

[0189] wherein, is the underwater clarity index, is the underwater contrast index, and are the weight factors of the two sub-indices, respectively;

[0190] and The values of and are calculated by using the Turbid subset of the real underwater dataset UID2021 with MOS values related to turbidity changes, and the characteristics of all images are calculated and , and then the weights and are determined by multiple linear regression, and then the UICS is calculated.

[0191] S5, based on the four-option forced choice 4AFC method to determine the direction of the image, the correct probability of direction discrimination corresponding to different UICS values is counted, and the relationship curve between UICS and correct probability is fitted by Weibull psychometric function, as follows:

[0192] S51: The obtained images are discriminated by the four-option forced choice 4AFC method, and the four directions of "up, down, left and right" are pre-set, and the optimal judgment result is forcedly selected;

[0193] S52: For a triangle target with a fixed size , the direction correct judgment probability under different UICS (instead of traditional contrast ) is counted by changing the turbidity of the water body and repeating the 4AFC test many times; ;

[0194] S53: Weibull psychometric function is used for data fitting to establish the quantitative relationship between and UICS, and the specific calculation formula is as follows:

[0195] ;

[0196] wherein, is the correct judgment deviation rate, representing the probability of error operation or failure to accurately observe the actual situation under the condition of correct judgment, and is usually set to ; This refers to the probability of guessing correctly when the direction is difficult to determine; in the UTOD model, since one of the four preset directions "up, down, left, and right" must be selected each time, The value is fixed at 0.25; This is the threshold value for the UICS indicator, corresponding to a 75% probability of correct judgment. The slope of the curve reflects the system's sensitivity to differences in underwater image information; for example... Figure 6 As shown in (a) in the figure, the UICS value corresponding to a 75% correct judgment probability can be obtained through this curve.

[0197] S6. Repeat the test for multiple triangular targets of different sizes to generate UTOD curves. The specific steps are as follows:

[0198] By fitting the experimental data using the nonlinear least squares method, the parameters can be determined. and The optimal solution is found, thus successfully extracting the UICS index threshold. Repeat the above process for multiple different sizes, ultimately generating a graph with the spatial frequency of the triangle's angles as the x-axis. The UTOD curve with the ordinate as the vertical axis, such as Figure 6 As shown in (b);

[0199] This curve quantitatively demonstrates the resolution limits of the imaging system in an underwater environment, assuming consistent dimensions. The smaller the value, the stronger the imaging system's ability to perceive and distinguish underwater image information, and the better its imaging performance.

[0200] Compare the target imaging effects under different algorithms (e.g.) Figure 7 ) and UTOD curves of imaging results from different algorithms (e.g. Figure 8 The present invention provides higher clarity and contrast in target imaging under different turbidity levels, and the UTOD curve is also more accurate.

[0201] It is worth noting that all contents not described in detail in this invention are existing technologies and are well known to those skilled in the art.

[0202] Therefore, this invention provides an underwater imaging quality assessment method based on UTOD (Underwater Target Occurrence Discrimination). It integrates HVS (High-Speed ​​Detection) characteristics and corner-edge joint features to construct a comprehensive contrast-sharpness index (UICS), achieving automated target detection and orientation recognition, and improving the consistency between subjective and objective assessments. Experimental results show that the proposed assessment method is highly consistent with subjective scoring, exhibiting superior accuracy, stability, and objectivity compared to traditional indicators.

[0203] It should be pointed out finally that the above examples are only used to illustrate the technical solutions of the present application but not to limit it, and although the present application has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solutions of the present application can still be modified or replaced equivalently, and these modifications or equivalent replacements should not make the modified technical solutions deviate from the spirit and scope of the technical solutions of the present application.

Claims

1. A UTOD-based underwater imaging quality evaluation method, characterized in that, The method comprises the following steps: Step S1, collecting triangle target feature information, and processing the obtained image through a human eye visual system (HVS) simulation module; Step S2, constructing an underwater triangle direction discrimination threshold (UTOD) model, and using the UTOD model to discriminate the direction of the triangle target; Step S3, constructing a contrast-sharpness comprehensive index (UICS) including an underwater sharpness index and an underwater contrast index; Step S4, calculating the contrast-sharpness comprehensive index; Step S5, using a four-alternative forced choice (4AFC) method to discriminate the direction of the image, and after counting the direction discrimination correct probability corresponding to different UICS values, fitting the relationship curve between UICS and the correct probability through a psychometric function Weibull; Step S6, repeatedly testing a plurality of triangle targets of different sizes to generate a UTOD curve; In step S2, the specific steps are as follows: Step S21, determining the pointing direction of the triangle target, calculating the geometric center based on the position information of the three vertices, and the specific calculation formula is as follows: ； wherein is the geometric center of the triangle, is the horizontal coordinate of the triangle vertex, is the vertical coordinate of the triangle vertex, is the number of triangle vertices; Step S22, taking the barycenter as the reference, the triangle is divided into two parts in horizontal and vertical directions, and the corresponding areas are denoted as , and , , the ratio of the areas of the parts is calculated to obtain the area factor , The specific calculation formula is as follows: ； wherein, is an area factor for the upper part, is an area factor for the left part, is an area of the upper half of the triangular figure, is an area of the lower half of the triangular figure, is an area of the left half of the triangular figure, is an area of the right half of the triangular figure; Step S23, judging the direction of the triangle target according to the judgment rule, and the specific judgment rule is as follows: ； wherein is a tolerance value; In step S3, the contrast-sharpness comprehensive index (UICS) is constructed, and the specific steps are as follows: Step S31, calculating the underwater sharpness index, and the steps are as follows: Step S311, applying to the image Operators, calculating the horizontal and vertical direction gradient, get the edge strength map, the formula as follows: ； wherein, is a gradient in the horizontal direction, is a gradient in the horizontal direction operator, is the original target image, is a gradient in the vertical direction, is a gradient in the vertical direction operator; Step S312, defining the calculated edge intensity map as the gradient amplitude, and the formula is as follows: ； wherein is an edge strength image; Step S313, multiplying the edge intensity map and the original image pixel by pixel to retain the intensity information of the edge region, and the formula is as follows: ； wherein is the edge intensity weighted gray scale enhanced image; Step S314, calculating the relative contrast based on the blocks to obtain , the formula is as follows: ； ； wherein, is an underwater sharpness index, is an edge map energy, is a number of blocks in the vertical direction, is a number of blocks in the horizontal direction, is a vertical block index, is a horizontal block index, is a maximum gray value within a row, column block, is a minimum gray value within a row, column block; Step S32, calculating the underwater contrast index The formula is as follows: ； ； wherein, is an underwater contrast index, is an average Michelson contrast in a local region of the image, is a multiplicative aggregation operation under PLIP, is a difference operation under PLIP, is an additive aggregation operation under PLIP; In step S4, the formula is as follows: ； wherein is a contrast-sharpness composite index, and is a weight factor; In step S5, the four-alternative forced choice (4AFC) method is used to discriminate the direction of the image, and the direction discrimination correct probability corresponding to different UICS values is counted, and the formula is as follows: ； wherein, Pd is the probability of correct direction determination under different UICS, Pb is the bias value of correct determination, Pc is the probability of random guess when the direction is difficult to distinguish, Pth is the UICS index threshold value, Pc is the curve slope. 2.The UTOD-based underwater imaging quality evaluation method according to claim 1, characterized in that, In step S1, the specific steps are as follows: Step S11, processing the image by using the human eye visual system (HVS) simulation module, and the specific steps are as follows: Step S111, simulating visual nonlinear effects, performing nonlinear luminance response conversion based on the Weber-Fechner law, and the calculation formula is as follows: ； wherein, is the perceived luminance, is a constant related to the average luminance of the whole image, is the objective luminance, is a basic response threshold value that characterizes the visual system; Step S112, simulating a multi-channel structure, and performing multi-channel modulation effect simulation on the signal by using a band-pass filter with specific spatial frequency and direction characteristics, and the specific steps are as follows: Gabor filter set is used to extract directional texture features, and the specific calculation formula is as follows: ； ； ； Gabor filter with direction is is the image pixel coordinate, is the image pixel coordinate after coordinate transformation, is the filter direction, is the center frequency, is the standard deviation; Four-channel filtering is performed on the input image, and the specific calculation formula is as follows: ； wherein, is the convolution result of the input underwater image with a Gabor filter whose direction is . Key information of the image in each direction is extracted, and the energy of each direction is extracted, and the specific calculation formula is as follows: ； wherein is energy values in the direction, is the total number of image pixels; Step S113, weighting and fusing the directional subbands by using a contrast sensitivity function (CSF), and the specific steps are as follows: The contrast sensitivity equation (CSF) is used to describe the correlation between the perception of the human eye visual system to the image contrast and the background brightness and spatial frequency, and the specific calculation formula is as follows: ； ； wherein is a contrast sensitivity function value, is a frequency component in the horizontal direction of the image, is a frequency component in the vertical direction, is a spatial frequency; Each directional subband is weighted, and the specific calculation formula is as follows: ； wherein is energy values in the CSF weighted direction; Each directional feature map is fused to obtain an HVS energy feature image, and the specific calculation formula is as follows: ； wherein, is the HVS energy feature image; Step S12, obtaining triangle target features based on a corner-edge joint feature extraction target detection method, and using an improved Harris corner detection strategy, and the specific steps are as follows: Step S121, detection is performed according to the characteristics of the triangle vertex, and the specific calculation formula is as follows: ； ； wherein, is the determinant of the matrix , is the trace of the matrix , is an empirical constant, is the gradient in the direction, is the gradient in the direction, is the Harris corner response function; Step S122, the obtained numerous corner points are screened, and the specific screening rule is: and the first 3 largest response points are retained; wherein, is a set dynamic threshold value, while the first 3 largest response points are forced to be reserved; Step S13, the target detection method based on the corner-edge joint feature extraction obtains the triangle target feature, optimizes the Canny edge detection, and the specific steps are as follows: Step S131, according to the actual underwater imaging environment, a double threshold rule is set, and the specific calculation formula is as follows: ； wherein, is a maximum value of the gradient in the image, is a low threshold value, is a high threshold value; Step S132, based on the characteristics that the real corner points are located on or close to the edge, the spatial relationship between the edge points and the corner points is used to eliminate the false corner points, and for each Harris corner point, the minimum Euclidean distance between the Harris corner point and all Canny edge points is calculated: ； wherein, is the minimum Euclidean distance between each Harris corner point and all Canny edge points, is the corner coordinate, is the edge point coordinate, denotes the minimum value among all edge point coordinates . 3.The underwater imaging quality evaluation method based on UTOD according to claim 1, characterized in that, In step S6, repeated tests are performed on multiple triangular targets of different sizes to generate a UTOD curve, specifically: using a nonlinear least squares method to fit the experimental data to determine the parameters With the optimal solution, the UICS index threshold is extracted Finally, a UTOD curve is generated with the triangular corner spatial frequency as the horizontal coordinate, and the vertical coordinate.

Images