A ship function damage assessment method based on feature mapping cloud model

Abstract

The present invention discloses a ship function damage assessment method based on a feature mapping cloud model, aiming to improve the accuracy and reliability of damage assessment. The method comprises the following steps: first, the characteristic parameters of the ship damage area are extracted by using an image threshold segmentation method and a spectral residual method; second, a damage cloud model is constructed by an inverse cloud generator MBCT-SR algorithm, and a mapping relationship between characteristic parameters and damage degree is established; then, effective assessment characteristic parameters are selected according to the damage cloud model; finally, the probability of damage degree of each functional area of ​​the ship is calculated by combining the primary assessment characteristics and the secondary assessment characteristic values, so as to realize the determination of damage degree of the functional area of ​​the ship. The method of the present invention has high accuracy, strong robustness and wide applicability, can effectively assess the degree of damage of ship functions in complex environments, and provide a scientific basis for ship damage control and recovery.

Description

Technical Field

[0001] The present invention relates to the field of sea area monitoring, and in particular to a ship function damage assessment method based on a feature mapping cloud model. Background Art

[0002] The damage assessment of target objects aims to use remote sensing images to accurately detect the damaged areas of targets caused by disasters or accidental attacks on civilian or military vessels in surveillance scenes through comprehensive image analysis and intelligent reasoning, and quantitatively analyze the degree of functional damage of the objects through damage models, thereby providing quantitative auxiliary interpretation technology for maritime surveillance and objective definition basis for ship maritime operation safety, maintenance, commercial claims, etc. At present, traditional ship damage assessment methods mainly rely on empirical judgment and simple mathematical models, and have problems such as strong subjectivity, poor quantification, and weak reliability. In particular, traditional methods are difficult to effectively deal with the influence of multiple factors in complex environments.

[0003] In order to solve the above problems, a ship function damage assessment method based on feature mapping cloud model is proposed. This method can achieve quantitative assessment of the damage degree of ship functional area through feature extraction of ship damage area and cloud model construction, with high accuracy and strong robustness. Summary of the invention

[0004] The purpose of the present invention is to provide a ship function damage assessment method based on a feature mapping cloud model.

[0005] To achieve the above object, the present invention is implemented according to the following technical solutions:

[0006] The present invention comprises the following steps:

[0007] S1: Feature extraction of ship damage area: The image threshold segmentation method is used to extract the ship damage area, and the spectral residual method is used to extract the damage features of the ship damage area;

[0008] S2: Ship damage cloud model construction: Through target sample training, the reverse cloud generator MBCT-SR algorithm is used to establish a mapping curve trend chart between the characteristic parameters of different targets and their functional damage degree;

[0009] S3: Selection of effective evaluation characteristic parameters: Select effective evaluation characteristic parameters based on the trend of the ship damage cloud model mapping relationship curve;

[0010] S4: Determination of damage degree of ship functions: Determination of damage degree of ship functions is achieved based on effective fusion feature reasoning based on the membership of ship damage cloud model.

[0011] The damage characteristics of the damaged area of ​​the ship in step S1 include eight characteristics, namely, damage area ratio, damage relative perimeter, damage density, average brightness change rate, contrast change rate, correlation change rate, energy change rate, and consistency change rate, to characterize different degrees of damage to the ship object.

[0012] The step S3 specifically comprises the following steps:

[0013] S31: Significant feature selection method: observe whether the feature cloud moves in the same direction when the degree of regional damage increases;

[0014] S32: Determination of the main evaluation feature: If the feature cloud graph generally shows an increasing or decreasing trend, then this feature is the main evaluation feature;

[0015] S33: Determination of secondary assessment features: If the feature cloud image generally moves in the same direction but is confused, then the feature is a secondary assessment feature. The secondary assessment feature participates in the assessment of damage situations that are not affected by confusion.

[0016] S34: Conditions for feature participation in assessment: As the degree of regional damage increases, features of the cloud image that do not move in the same direction as a whole do not participate in the assessment. In the island area damage assessment, the damage density feature participates or does not participate in the assessment under specific conditions: moderate to severe damage or severe damage, as well as moderate damage or complete damage: the damage density feature does not participate in the assessment; slight damage: the damage density feature participates in the assessment.

[0017] The step S4 specifically comprises the following steps:

[0018] S41: Initial damage degree determination: Determine the initial damage degree of the functional area based on the membership set of the main assessment features;

[0019] S42: Sub-assessment feature verification: Use sub-assessment features to verify the initial damage level to ensure the accuracy of the judgment; in specific damage situations, eliminate the influence of confusing features to ensure the reliability of the assessment results;

[0020] S43: Final damage degree judgment: Determine the final damage degree of the functional area of ​​the ship by integrating the results of the primary and secondary assessment characteristics; ensure that the assessment results are consistent with the actual damage situation and improve the accuracy and effectiveness of the assessment.

[0021] The beneficial effects of the present invention are:

[0022] The present invention is a ship function damage assessment method based on a feature mapping cloud model. Compared with the prior art, the present invention has the following technical effects:

[0023] Improve assessment accuracy: Establish a damage cloud model through the reverse cloud generator MBCT-SR algorithm to achieve accurate mapping of characteristic parameters and functional damage degree.

[0024] Enhanced assessment reliability: The spectral residual method is used to extract ship damage characteristic parameters, reduce noise interference, and enhance the robustness of the assessment results.

[0025] Multi-feature fusion assessment: By selecting primary and secondary assessment features, multiple damage feature parameters are comprehensively considered to improve the comprehensiveness and reliability of the assessment.

[0026] Wide applicability: This method is not only applicable to damage assessment of civilian ships, but also can be extended to damage assessment of military ships. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 It is a block diagram of ship function damage assessment of feature mapping cloud model;

[0028] Figure 2 It is a flow chart of mapping characteristic parameters and damage degree;

[0029] Figure 3 These are examples of damage images of typical functional areas of ships, including damage illustrations of civilian ships and naval vessels (the red frame indicates the significant damage area);

[0030] Figure 4 is the characteristic cloud model of the island area of ​​the ship, which are the corresponding relationship curves of the (8) characteristic change values ​​of the island functional area and the five levels of damage degree (determined value);

[0031] Figure 5 is the characteristic cloud model of the flight deck elevator area of ​​the ship, which are the corresponding relationship curves of the (8) characteristic change values ​​of the flight deck elevator functional area and the five-level damage degree (determined value);

[0032] Figure 6 It is the damage model assessment result for the damaged ship image example of "moderate damage to the island + slight damage to the flight deck", showing five types of damage degree values ​​for the ship's functional areas (the color bar represents its certainty probability value). DETAILED DESCRIPTION

[0033] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments. The illustrative embodiments and descriptions of the present invention are used to explain the present invention but are not intended to limit the present invention.

[0034] like Figure 1 and 2 As shown: The present invention comprises the following steps:

[0035] S1: Ship damage area feature extraction: The ship damage area is extracted using the image threshold segmentation method, and the damage features of the ship damage area are extracted using the spectral residual method; the damage features of the ship damage area include damage area ratio, damage relative perimeter, damage density, average brightness change rate, contrast change rate, correlation change rate, energy change rate, and consistency change rate to characterize the different damage degrees of the ship object. The parameters of the 8 features are calculated as follows:

[0036]

[0037] Where ratio_area is the ratio of the damaged area of ​​the corresponding functional area; size(x) indicates the size of x; find(label==n) indicates the pixel position in label whose pixel value is equal to n; label is the labeled image; damage is the pixel set of the damaged area; n indicates the functional area of ​​the damaged area; C is the relative perimeter of the damaged area; B is the perimeter of the corresponding functional area damaged, that is, the number of pixels in contact between the damaged part and the non-damaged part of each functional area; L is the length of the long axis of the entire ship; M is the density parameter of the corresponding functional damaged area; ratio_L is the average brightness change rate of the corresponding functional area of ​​the ship; L_before and L_after respectively indicate the brightness before and after damage. , and the average brightness of the corresponding functional areas of the ship before and after damage; ratio_D, ratio_X, ratio_N, ratio_Y are respectively the contrast change rate, correlation change rate, energy change rate and consistency change rate of the corresponding functional areas of the ship; D, X, N, Y are respectively the contrast, correlation, energy and consistency of the corresponding functional areas of the ship; D_before, X_before, N_before, Y_before are respectively the contrast, correlation, energy and consistency of the corresponding functional areas of the ship before damage; D_after, X_after, N_after, Y_after are respectively the contrast, correlation, energy and consistency of the corresponding functional areas of the ship after damage.

[0038] S2: Ship damage cloud model construction: Through target sample training, the reverse cloud generator MBCT-SR algorithm is used to establish a mapping curve trend chart between the characteristic parameters of different targets and their functional damage degree; the MBCT-SR algorithm cloud model is constructed as follows:

[0039]

[0040] In the formula, is the cloud droplet expectation; N is the number of cloud droplets; x i For each cloud drop data. is the average number of randomly sampled cloud droplets; r is the number of randomly sampled cloud droplets; x ij is the randomly sampled cloud droplet data.i 2 is the sampled cloud droplet variance; is the entropy of the cloud droplets. 2 Y i 2 Expectations; DY 2 Y i 2 The variance of . It is the cloud droplet super entropy.

[0041] S3: Selection of effective evaluation characteristic parameters: Select effective evaluation characteristic parameters according to the trend of the ship damage cloud model mapping relationship curve; specifically including the following steps:

[0042] The method for distinguishing whether a feature is significant in this scheme is: as the degree of regional damage increases, whether the newly generated clouds of this feature move in the same direction in the cloud map. If so, it means that as the degree of regional damage increases, the feature generally shows an increasing or decreasing trend, and can be used as the basis for damage assessment of the functional area. This feature is the main assessment feature; if it moves in the same direction as a whole, but clouds with similar degrees of damage are confused, then as a secondary assessment feature, it participates in the assessment transformation in the scenario of partial damage in the functional area; if as the degree of regional damage increases, the cloud map does not move in the same direction as a whole, then this feature does not participate in the assessment. Among them, the secondary assessment feature only participates in the assessment of the partial damage that is not affected by confusion. For example, when transforming the damage assessment of the island area, the damage density feature will confuse the heavy damage and severe damage in the island. Then, under its influence, when the main assessment feature considers that the island area is moderately or severely damaged or severely damaged, and its adjacent damage degree is moderately damaged or completely damaged, the damage density feature does not participate in the damage assessment transformation of the island area; only when the main assessment feature considers that the island area is slightly damaged, the damage density feature participates in the damage assessment transformation of the island area.

[0043] S4: Ship function damage degree judgment: Based on the effective fusion feature reasoning of the ship damage cloud model membership, the ship function damage degree judgment is realized. Specifically, the following steps are included:

[0044] First, for the main assessment feature corresponding to the functional area, the main assessment feature certainty set is determined through the cloud model fitting function, and the main assessment feature certainty is normalized as the probability array of the main assessment feature for assessing the damage degree of the functional area. The main assessment feature certainty formula is as follows:

[0045] Z i =(Z 0,i ,…,Z m,i )

[0046] In the formula, Z i is the i-th main evaluation feature certainty set; Z 0,i,…,Z m,i is the degree of certainty from zero damage to complete damage of the functional area in the i-th main evaluation characteristic cloud model; m is the number of damage degrees designed for the functional area.

[0047] Normalize the main evaluation feature certainty, and the formula is as follows:

[0048]

[0049] Where D i is the normalized result of the determination degree of the i-th main evaluation feature.

[0050] In the case of only one main assessment feature, the damage degree with the highest certainty in the cloud model corresponding to the main assessment feature data is used as the initial damage degree of the region to determine the secondary assessment feature; if there is more than one main assessment feature, it is necessary to first fuse the main assessment features, and then determine the damage degree with the highest probability as the initial damage degree of the region through the fusion result to determine the secondary assessment feature. The main assessment feature fusion is as follows:

[0051]

[0052] Where D is the fusion result of the main evaluation features; n is the number of main evaluation features.

[0053] After determining the secondary assessment feature, it is necessary to convert the feature data into a probability array of damage degree. The secondary assessment feature certainty set is determined by the cloud model fitting function, and the secondary assessment feature certainty is normalized to serve as the probability array of the secondary assessment feature for the damage degree assessment of the functional area. Finally, the primary and secondary assessment features are fused based on the normalized results of the primary and secondary assessment feature certainty to determine the probability array of the damage degree of the functional area. The primary and secondary assessment features are fused as follows:

[0054]

[0055] Where A is the fusion result of primary and secondary assessment features, that is, the probability array of functional area damage degree; n is the number of primary assessment features; m is the number of secondary assessment features; C j Normalized result of feature certainty of the j-th evaluation.

[0056] by Figure 3 For example, a sample image of damage to a ship or vessel; Figure 4 , Figure 5 Example Figure 3 (b) The characteristic cloud model curve of the ship's island and flight deck functional area reflects the quantitative relationship between the change of graphic characteristics and the damage degree evaluation; according to the primary and secondary evaluation feature fusion damage model proposed by this method, Figure 6The damage probability values ​​of the main functional areas of the ship are displayed (5 levels: zero damage, slight damage, moderate damage, moderately severe damage, and complete damage). The moderate damage probability value of the island is the largest, and the other functional areas are all zero damage or slight damage, so it is judged as moderate damage. Therefore, its judgment is accurate and reasonable.

[0057] The technical solution of the present invention is not limited to the above-mentioned specific embodiments. All technical variations made according to the technical solution of the present invention fall within the protection scope of the present invention.

Claims

1. A ship function damage assessment method based on a feature mapping cloud model, characterized in that: The following steps are involved: S1: Feature extraction of ship damage area: The image threshold segmentation method is used to extract the ship damage area, and the spectral residual method is used to extract the damage features of the ship damage area; S2: Ship damage cloud model construction: Through target sample training, the reverse cloud generator MBCT-SR algorithm is used to establish a mapping curve trend chart between the characteristic parameters of different targets and their functional damage degree; S3: Selection of effective evaluation characteristic parameters: Select effective evaluation characteristic parameters based on the trend of the ship damage cloud model mapping relationship curve; S31: Significant feature selection method: observe whether the feature cloud moves in the same direction when the degree of regional damage increases; S32: Determination of the main evaluation feature: If the feature cloud graph generally shows an increasing or decreasing trend, then this feature is the main evaluation feature; S33: Determination of secondary assessment features: If the feature cloud image generally moves in the same direction but is confused, then the feature is a secondary assessment feature. The secondary assessment feature participates in the assessment of damage situations that are not affected by confusion. S34: Conditions for feature participation in assessment: As the damage level of the area increases, features that do not move in the same direction as a whole in the cloud map do not participate in the assessment. In the damage assessment of the island area, the damage density feature participates in or does not participate in the assessment under certain conditions: moderate or severe damage, as well as moderate or complete damage: the damage density feature does not participate in the assessment; mild damage: the damage density feature participates in the assessment; S4: Determination of damage degree of ship functions: Determination of damage degree of ship functions is realized based on effective feature fusion reasoning of membership of ship damage cloud model.

2. The ship function damage assessment method based on the feature mapping cloud model according to claim 1 is characterized in that: The damage characteristics of the damaged area of ​​the ship in step S1 include eight characteristics, namely, damage area ratio, damage relative perimeter, damage density, average brightness change rate, contrast change rate, correlation change rate, energy change rate, and consistency change rate, to characterize different degrees of damage to the ship object.

3. The ship function damage assessment method based on the feature mapping cloud model according to claim 2 is characterized in that: The parameters of the 8 features are calculated as follows: Where ratio_area is the ratio of the damaged area of ​​the corresponding functional area; size(x) indicates the size of x; find(label==n) indicates the pixel position in label whose pixel value is equal to n; label is the labeled image; damage is the pixel set of the damaged area; n indicates the functional area of ​​the damaged area; C is the relative perimeter of the damaged area; B is the perimeter of the corresponding functional area damaged, that is, the number of pixels in contact between the damaged part and the non-damaged part of each functional area; L is the length of the long axis of the entire ship; M is the density parameter of the corresponding functional damaged area; ratio_L is the average brightness change rate of the corresponding functional area of ​​the ship; L_before and L_after respectively indicate the brightness before and after damage. , and the average brightness of the corresponding functional areas of the ship before and after damage; ratio_D, ratio_X, ratio_N, ratio_Y are respectively the contrast change rate, correlation change rate, energy change rate and consistency change rate of the corresponding functional areas of the ship; D, X, N, Y are respectively the contrast, correlation, energy and consistency of the corresponding functional areas of the ship; D_before, X_before, N_before, Y_before are respectively the contrast, correlation, energy and consistency of the corresponding functional areas of the ship before damage; D_after, X_after, N_after, Y_after are respectively the contrast, correlation, energy and consistency of the corresponding functional areas of the ship after damage.

4. The ship function damage assessment method based on the feature mapping cloud model according to claim 1 is characterized in that: The reverse cloud generator MBCT-SR algorithm in step S2 is as follows: In the formula, is the cloud droplet expectation; N is the number of cloud droplets; x i For each cloud drop data; is the average number of randomly sampled cloud droplets; r is the number of randomly sampled cloud droplets; x ij is the randomly sampled cloud droplet data; Y i 2 is the sampled cloud droplet variance; is the entropy of cloud droplets; EY 2 Y i 2 Expectations; DY 2 Y i 2 The variance of is the cloud droplet super entropy.

5. The ship function damage assessment method based on the feature mapping cloud model according to claim 1 is characterized in that: The step S4 specifically comprises the following steps: S41: Initial damage degree determination: Determine the initial damage degree of the functional area based on the membership set of the main assessment features; S42: Sub-assessment feature verification: Use sub-assessment features to verify the initial damage level to ensure the accuracy of the judgment; in specific damage situations, eliminate the influence of confusing features to ensure the reliability of the assessment results; S43: Final damage degree judgment: Determine the final damage degree of the functional area of ​​the ship by integrating the results of the primary and secondary assessment characteristics; ensure that the assessment results are consistent with the actual damage situation and improve the accuracy and effectiveness of the assessment.

Images