Building structure stability detection method based on multispectral image

Abstract

The invention relates to the technical field of image processing, in particular to a building structure stability detection method based on a multispectral image, and the method comprises the steps: determining an image state according to a background coefficient and a building compactness coefficient, and determining a processing mode according to the image state, determining a building area according to the feature line; and when a building area is determined according to the special point clustering area, determining a sampling point window area according to the sampling point mutation coefficient of the division unit, and performing window analysis on each sampling point in the first spectral image by adopting the determined sampling point window area to determine a special point, determining whether the special point clustering area is a building area according to the brightness difference coefficient and the number of the special points; determining an analysis mode according to the image difference degree and the building area evaluation coefficient; according to the method, the accuracy degree of building structure stability prediction can be improved while the accuracy degree of building area identification in the multispectral image is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to a building structure stability detection method based on multispectral images. Background Art

[0002] After an earthquake, a rapid and accurate assessment of the stability of building structures is an important basis for formulating rescue plans and post-disaster reconstruction plans. With the development of remote sensing technology, key information on building stability and damage is often extracted through multispectral images. However, in practical applications, due to the obstruction and confusion of trees and roads, as well as the density of building areas, the recognition effect of building areas in multispectral images is poor. Therefore, how to improve the accuracy of building area recognition in multispectral images to improve the accuracy of building structure stability prediction is a technical problem that needs to be urgently solved by technical personnel in this field.

[0003] Chinese Patent Publication No. CN103077515B discloses a method for detecting building changes in multispectral images, including: first, performing change detection based on the pixel ratio method to obtain the change area of ​​all ground objects, and this area is used as a candidate area for building changes. Change detection based on building features is performed on the change area detected by the ratio method. Since the use of one feature is prone to false detection, this method uses a combination of texture features and hue features to perform feature-level change detection. It can be seen that the above technical solution has the following problems: the building area is determined only based on the texture feature and hue feature sequence, and the recognition method is not adaptively adjusted according to the actual state of the multispectral image, resulting in poor recognition accuracy of the building area. Summary of the invention

[0004] To this end, the present invention provides a building structure stability detection method based on multispectral images, so as to overcome the problem that the existing technology has a single building area recognition method and fails to adaptively adjust the recognition method according to the actual state of the multispectral image, resulting in poor building area recognition accuracy.

[0005] To achieve the above object, the present invention provides a method for detecting building structure stability based on multispectral images, comprising:

[0006] Acquire spectral image information, determine the image state according to the background coefficient and the building density coefficient, and determine the processing method according to the image state, the processing method is to determine the building area according to the special point clustering area, or to determine the building area according to the characteristic line;

[0007] When determining the building area according to the special point cluster area, determine the image division method according to the sampling point disorder coefficient and the characteristic line turbulence amplitude to obtain a number of division units, and determine the sampling point window area according to the sampling point mutation coefficient of the division unit, use the determined sampling point window area to perform window analysis on each sampling point in the first spectral image to determine the special point, and determine whether the special point cluster area is a building area according to the brightness difference coefficient and the number of special points;

[0008] When determining the building area based on feature lines, the building area is determined based on the closed area formed by the feature lines;

[0009] When the building area determination is completed, an analysis method is determined according to the image difference between the second spectral image and the first spectral image and the building area assessment coefficient of the first spectral image. The analysis method is to determine the stability determination method according to the comparison area difference and the difference area mapping coefficient or to determine the stability according to the damage threshold of the building area;

[0010] The stability determination method is to determine the stability according to the difference of the comparison area or to determine the stability according to the influence coefficient of the adjacent area and the difference of the comparison area;

[0011] The image division method is to divide the image evenly according to the disorder threshold or to divide the image according to the position of the feature line.

[0012] Furthermore, the image state is determined according to the background coefficient and the building density coefficient, and the image state includes:

[0013] A first image state in which a background coefficient is greater than or equal to a preset background coefficient or a building density coefficient is greater than or equal to a preset building density coefficient;

[0014] A second image state in which the background coefficient is less than a preset background coefficient and the building density coefficient is less than a preset building density coefficient.

[0015] Further, a processing method is determined according to the image state, wherein:

[0016] In the first image state, the processing method is to determine the building area based on the special point cluster area;

[0017] In the second image state, the processing method is to determine the building area according to the feature lines.

[0018] Furthermore, the image division method is determined according to the sampling point disorder coefficient and the characteristic line turbulence amplitude to obtain a number of division units, wherein:

[0019] If the sampling point disorder coefficient is greater than or equal to the preset sampling point disorder coefficient or the characteristic line turbulence amplitude is greater than or equal to the preset characteristic line turbulence amplitude, the image division method is to divide the image uniformly according to the disorder threshold;

[0020] If the sampling point disorder coefficient is less than the preset sampling point disorder coefficient and the characteristic line turbulence amplitude is less than the preset characteristic line turbulence amplitude, the image division method is to divide according to the characteristic line position.

[0021] Furthermore, the sampling point window area is determined according to the sampling point mutation coefficient of the division unit, wherein,

[0022] The sampling point window area corresponding to a single sampling point is negatively correlated with the sampling point mutation coefficient corresponding to the division unit where the sampling point is located;

[0023] The sampling point window corresponding to a single sampling point is a square area centered on the sampling point.

[0024] Furthermore, a window analysis is performed on each sampling point in the first spectral image, wherein:

[0025] When performing window analysis on a single sampling point, the sampling point is recorded as the target sampling point;

[0026] If the window influence coefficient is greater than or equal to the preset window influence coefficient and the target sampling point satisfies the preset window state, then the target sampling point is a special point;

[0027] The preset window condition is that the color intensity value of the target sampling point is the maximum value among the color intensity values ​​corresponding to each sampling point in the sampling point window corresponding to the target sampling point.

[0028] Furthermore, the window influence coefficient corresponding to a single sampling point is determined according to whether the characteristic window line corresponding to the sampling point meets the preset placement conditions and the proportion of the same sampling points, wherein:

[0029] If the preset placement conditions are met or the proportion of the same sampling points is not within the preset range of the same sampling point proportion, the window influence coefficient is determined according to the specific index;

[0030] If the preset placement conditions are not met and the proportion of the same sampling points is within the preset range of the proportion of the same sampling points, the window influence coefficient is determined according to the angle disorder value of the same sampling points;

[0031] When determining the window influence coefficient according to the same sampling point angle disorder value, if the same sampling point angle disorder value is greater than or equal to the preset same sampling point angle disorder value, the window influence coefficient is determined according to the same sampling point angle disorder value; if the same sampling point angle disorder value is less than the preset same sampling point angle disorder value, the window influence coefficient is determined according to the regional disorder degree and the local similarity;

[0032] The preset placement condition is that the line segment threshold is a standard threshold and the window line position relationship is a cross relationship.

[0033] Furthermore, whether the special point clustering area is a building area is determined according to the brightness difference coefficient and the number of special points, wherein:

[0034] If the brightness difference coefficient is less than the preset brightness difference coefficient and the number of special points is greater than or equal to the preset number of special points, the special point clustering area is a building area;

[0035] The special point clustering area is determined according to the special point threshold similarity coefficient and the special point spacing.

[0036] Furthermore, the analysis method is determined according to the image difference and the building area evaluation coefficient, wherein:

[0037] If the image difference is greater than or equal to the preset image difference or the building area assessment coefficient is greater than or equal to the preset building area assessment coefficient, the analysis method is to determine the stability determination method according to the comparison area difference and the difference area mapping coefficient;

[0038] If the image difference is less than the preset image difference and the building area assessment coefficient is less than the preset building area assessment coefficient, the analysis method is to determine the stability according to the damage threshold of the building area;

[0039] The stability of a single building area is negatively correlated with the damage threshold of the building area.

[0040] Furthermore, the stability determination method is determined according to the comparison area difference and the difference area mapping coefficient, wherein:

[0041] If the comparison area difference is greater than or equal to the preset comparison area difference or the difference area mapping coefficient is greater than or equal to the preset difference area mapping coefficient, the stability is determined according to the comparison area difference;

[0042] If the comparison region difference is less than the preset comparison region difference and the difference region mapping coefficient is less than the preset difference region mapping coefficient, the stability is determined according to the neighboring region influence coefficient and the comparison region difference.

[0043] Compared with the prior art, the beneficial effect of the present invention lies in that, in the technical solution of the present invention, the image state is determined according to the background coefficient and the building density coefficient, and the background coefficient and the building density coefficient are used to effectively reflect the clarity of the building and the density of the building in the spectral image. Then, different processing methods are adaptively selected according to the image state, so that the selected processing method can accurately determine the building area, avoiding the problem of poor recognition accuracy caused by misjudging non-building areas as building areas, and also helping to reduce processing time and improve the overall processing efficiency of spectral images.

[0044] Furthermore, the present invention effectively reflects the regularity of building area distribution according to the sampling point disorder coefficient and the characteristic line turbulence amplitude, and then adaptively selects different image division methods according to the sampling point disorder coefficient and the characteristic line turbulence amplitude, so that the selection of image division method is more in line with the actual situation of the image, and can be applicable to various complex buildings and image types, avoiding the problem of poor special point recognition effect caused by unreasonable division of division units, thereby improving the recognition accuracy of building areas in spectral images.

[0045] Furthermore, the present invention performs window analysis on each sampling point in the first spectral image, so that the analysis range can be limited to the local area around the target point, reducing the overall calculation amount. The sampling point mutation coefficient of the division unit can effectively reflect the change of the sampling point in the division unit, and then the sampling point window area is determined according to the sampling point mutation coefficient of the division unit. The analysis range can be flexibly adjusted to adapt to the characteristics of buildings of different scales. At the same time, it can also reduce the influence of interference factors such as noise on the recognition results of the building area to a certain extent, thereby improving the accuracy of recognition.

[0046] Furthermore, in the present invention, whether the characteristic window line corresponding to the sampling point meets the preset placement conditions and the proportion of the same sampling points effectively reflects the inclination of the building in the sampling point window and the edge degree of the sampling point, and then determines the window influence coefficient according to whether the characteristic window line corresponding to the sampling point meets the preset placement conditions and the proportion of the same sampling points. Then, the window influence coefficient can effectively identify the special points representing the building area. When the inclination of the building is high, the disorder degree of the sampling points in the sampling point window is effectively reflected by the angle disorder value of the same sampling point, and then determines the window influence coefficient according to the angle disorder value of the same sampling point, so that the determination of the window influence coefficient is more in line with the actual application scenario, and the accuracy of building area recognition in complex environments is improved.

[0047] Furthermore, the present invention effectively reflects the degree of difference of sampling points in the special point clustering area through the brightness difference coefficient and the number of special points, and then determines whether the special point clustering area is a building area according to the brightness difference coefficient and the number of special points, so that the determination of the building area is more in line with the actual application scenario, reduces the problem of misjudging non-building areas as building areas, and improves the accuracy of building area identification.

[0048] Furthermore, the present invention effectively reflects the difference between the second spectral image and the first spectral image and the distribution of building areas through image difference and building area evaluation coefficient, and then adaptively selects different analysis methods according to the image difference between the second spectral image and the first spectral image and the building area evaluation coefficient of the first spectral image, so that the selection of analysis method can be adaptively adjusted according to the characteristics of the image, thereby improving the accuracy of the stability judgment of the building structure. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 It is a flow chart of the building structure stability detection method based on multi-spectral images of the present invention;

[0050] Figure 2 A flow chart of the present invention for determining a processing method according to an image state;

[0051] Figure 3 This is a flow chart of the present invention for determining an image division method based on a sampling point disorder coefficient and a characteristic line turbulence amplitude;

[0052] Figure 4 The present invention is a flow chart for determining an analysis method according to image difference and building area evaluation coefficient. DETAILED DESCRIPTION

[0053] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0054] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the protection scope of the present invention.

[0055] It should be noted that, in the description of the present invention, terms such as "up", "down", "left", "right", "inside" and "outside" indicating directions or positional relationships are based on the directions or positional relationships shown in the drawings. This is merely for the convenience of description and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it cannot be understood as a limitation on the present invention.

[0056] In addition, it should be noted that in the description of the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0057] See also Figures 1 to 4 As shown, the present invention provides a building structure stability detection method based on multispectral images, comprising:

[0058] Acquire spectral image information, determine the image state according to the background coefficient and the building density coefficient, and determine the processing method according to the image state, the processing method is to determine the building area according to the special point clustering area, or to determine the building area according to the characteristic line;

[0059] When determining the building area according to the special point cluster area, determine the image division method according to the sampling point disorder coefficient and the characteristic line turbulence amplitude to obtain a number of division units, and determine the sampling point window area according to the sampling point mutation coefficient of the division unit, use the determined sampling point window area to perform window analysis on each sampling point in the first spectral image to determine the special point, and determine whether the special point cluster area is a building area according to the brightness difference coefficient and the number of special points;

[0060] When determining the building area based on feature lines, the building area is determined based on the closed area formed by the feature lines;

[0061] When the building area determination is completed, an analysis method is determined according to the image difference between the second spectral image and the first spectral image and the building area assessment coefficient of the first spectral image. The analysis method is to determine the stability determination method according to the comparison area difference and the difference area mapping coefficient or to determine the stability according to the damage threshold of the building area;

[0062] The stability determination method is to determine the stability according to the difference of the comparison area or to determine the stability according to the influence coefficient of the adjacent area and the difference of the comparison area;

[0063] The image division method is to divide the image evenly according to the disorder threshold or to divide the image according to the position of the feature line.

[0064] The application scenario of the present invention is the detection of the stability of building structures after an earthquake. The spectral image information includes a first spectral image and a second spectral image. The first spectral image is a multispectral image taken by a drone at a certain position before the earthquake. The second spectral image is a multispectral image taken by a drone at the same shooting position as the first spectral image after the earthquake. The first spectral image and the second spectral image both contain a number of sampling points, and the positions of the sampling points are different. A single sampling point corresponds to a color intensity value, and the color intensity value corresponding to the single sampling point is greater than or equal to 0 and less than or equal to 255. This is content that is easy for a person skilled in the art to understand and will not be described in detail.

[0065] In the present invention, several historical records are correspondingly set up, and any historical record records the background coefficient, building tightness coefficient, environmental impact coefficient, synchronization coefficient, sampling point disorder coefficient, characteristic line turbulence amplitude and window influence coefficient, etc. in the historical process of at least one building structure stability detection, and each historical record corresponds to a qualified mark, which records whether the stability detection process meets the user's needs. The qualified mark can be recorded manually. It can be understood that the user can determine whether the stability detection process meets the needs based on self-set indicators. The self-set indicators can be but not limited to the misjudgment index, which will not be elaborated here. Among them, the misjudgment index is the number of times the stability of the building area is incorrectly judged.

[0066] Specifically, the image state is determined according to the background coefficient and the building density coefficient, and the image state includes:

[0067] A first image state in which a background coefficient is greater than or equal to a preset background coefficient or a building closeness coefficient is greater than or equal to a preset building closeness coefficient;

[0068] A second image state in which the background coefficient is less than a preset background coefficient and the building density coefficient is less than a preset building density coefficient.

[0069] Among them, the background coefficient is the standard deviation of the color intensity values ​​corresponding to each sampling point in the background area, and the background area is the area in the first spectral image that is not divided into the reference rectangle; the building closeness coefficient is the average value of the closeness distances corresponding to each reference rectangle. For a single reference rectangle, the reference rectangle is recorded as the first analysis rectangle, and the other reference rectangles outside the first analysis rectangle are recorded as the second reference rectangle. The minimum value of the shortest distance from the center position of the first reference rectangle to the center position of each second reference rectangle is recorded as the closeness distance; the center position of a single reference rectangle is the center of the circumscribed circle of the reference rectangle;

[0070] For a single sampling point, the sampling point is recorded as a target sampling point. If the minimum intensity difference between the target sampling point and the sampling points adjacent to the target sampling point is greater than a preset difference, the target sampling point is a mutation sampling point. The value of the preset difference can be set by the user according to the actual situation. The greater the user's demand for image processing accuracy, the smaller the preset difference value is. A preset difference value is provided. The preset difference is the minimum value of the minimum intensity difference between each mutation sampling point and the sampling points adjacent to the corresponding mutation sampling point in the historical records that can meet the user's needs. The minimum intensity difference is confirmed by recording the sampling points adjacent to the target sampling point as reference sampling points, recording the minimum value of the intensity difference corresponding to each reference sampling point as the minimum intensity difference, and the intensity difference corresponding to a single reference sampling point is the absolute value of the difference between the color intensity value corresponding to the reference sampling point and the color intensity value corresponding to the target sampling point;

[0071] Perform association analysis on each mutation sampling point. When performing association analysis on a single mutation sampling point, record the mutation sampling point as the first mutation sampling point, record the mutation sampling points other than the first mutation sampling point as the first reference mutation sampling point, record the mutation sampling points whose interval distance from the first mutation sampling point is less than the preset interval distance and the set of the first mutation sampling point as an association combination, and continue to perform association analysis on the mutation sampling points not recorded in the association combination until all mutation sampling points are recorded in the association combination; record the minimum rectangle that can contain all mutation sampling points in a single association combination as a reference rectangle;

[0072] The values ​​of the preset background coefficient and the preset building closeness coefficient can be determined by the user according to the actual application scenario. The smaller the values ​​of the preset background coefficient and the preset building closeness coefficient, the greater the user's need to determine the building area based on the special point clustering area. A value of the preset background coefficient and the preset building closeness coefficient is provided, and the historical records of determining the building area based on the characteristic lines are detected. The average value of the background coefficient corresponding to the historical records that can meet the user's needs is recorded as the preset background coefficient, and the average value of the building closeness coefficient corresponding to the historical records that can meet the user's needs is recorded as the preset building closeness coefficient.

[0073] Specifically, the processing method is determined according to the image state, wherein:

[0074] In the first image state, the processing method is to determine the building area based on the special point cluster area;

[0075] In the second image state, the processing method is to determine the building area according to the feature lines.

[0076] When determining the building area according to the characteristic lines, the building area is determined according to the closed area composed of the characteristic lines, wherein the closed area composed of the characteristic lines is confirmed by identifying each characteristic line in the first spectral image to obtain a plurality of closed areas, and if each side of a single closed area is a characteristic line, then the closed area is a building area;

[0077] A characteristic line is a line segment that meets the preset characteristic conditions and all the sampling points in the line segment are mutation sampling points. The preset characteristic condition is that the external sampling points corresponding to the line segment are not mutation sampling points. The method for confirming the external sampling points is that for a line segment, the line segment is recorded as a target line segment, each sampling point in the target line segment is recorded as a line segment sampling point, each sampling point in the straight line where the target line segment is located except the line segment sampling point is recorded as a straight line sampling point, and the straight line sampling points adjacent to the line segment sampling points at both ends of the target line segment are recorded as external sampling points.

[0078] Specifically, the image division method is determined according to the sampling point disorder coefficient and the characteristic line turbulence amplitude to obtain a number of division units, where:

[0079] If the sampling point disorder coefficient is greater than or equal to the preset sampling point disorder coefficient or the characteristic line turbulence amplitude is greater than or equal to the preset characteristic line turbulence amplitude, the image division method is to divide the image uniformly according to the disorder threshold;

[0080] If the sampling point disorder coefficient is less than the preset sampling point disorder coefficient and the characteristic line turbulence amplitude is less than the preset characteristic line turbulence amplitude, the image division method is to divide according to the characteristic line position.

[0081] Among them, the method for confirming the turbulence coefficient of the sampling point is:

[0082] If the environmental impact coefficient is greater than or equal to the preset environmental impact coefficient or the synchronization coefficient is less than the preset synchronization coefficient, the sampling point disorder coefficient is determined according to the impact threshold and the sampling point radiation value, wherein the sampling point disorder coefficient = impact threshold + sampling point radiation value;

[0083] If the environmental impact coefficient is less than the preset environmental impact coefficient and the synchronization coefficient is greater than or equal to the preset synchronization coefficient, the sampling point disorder coefficient is determined according to the sampling point radiation value, wherein the sampling point disorder coefficient is positively correlated with the sampling point radiation value;

[0084] Impact threshold = environmental impact coefficient - synchronization coefficient, sampling point radiation value = number of mutation sampling points in the first spectral image + impact sampling point distribution coefficient, the impact sampling point distribution coefficient is the average value of the distance coefficients corresponding to each impact sampling point in the first spectral image, for a single impact sampling point, the impact sampling point is recorded as the target sampling point, and the other impact sampling points except the impact sampling point are recorded as reference sampling points, and the average value of the shortest distance from the target sampling point to each reference sampling point is recorded as the distance coefficient corresponding to the target sampling point, and the impact sampling point is a mutation sampling point that is not on the characteristic line;

[0085] Environmental impact coefficient = illumination coefficient + attitude influence coefficient. It can be understood that excessive illumination intensity or excessive change in the attitude of the drone will affect the color intensity value of the sampling point in the image, thereby increasing the disorder of the sampling point. The present invention can monitor the flight trajectory of the drone in real time. The monitoring point in the present invention is set by the user. A method for setting a monitoring point is provided. Every 1s is recorded as a monitoring point, that is, the position of the drone is recorded once every 1s. The monitoring point where the drone flies to the position where the first spectral image is taken is recorded as the target monitoring point, and the monitoring point adjacent to the target monitoring point and earlier than the target monitoring point is recorded as the reference monitoring point. The drone position distribution corresponding to the target monitoring point and the reference monitoring point is recorded as the target position and the reference position. The calculation formula of the attitude influence coefficient ε is, is the vector corresponding to the target position, is the vector corresponding to the reference position, the vectors corresponding to the target position and the reference position are tangent to the moving trajectory and the direction of the vector is the same as the moving direction; the illumination coefficient is the illumination intensity of the target monitoring point when the UAV flies to the target monitoring point, and the illumination intensity is measured by a photoelectric sensor, which is easy for technicians in this field to understand and will not be described in detail;

[0086] Synchronization coefficient = 1 / channel time difference, where the channel time difference is the time interval between the earliest time point at which the spectral channel image is captured in each spectral channel and the earliest time point at which the spectral channel image is captured in each spectral channel. The spectral channels include but are not limited to red, green, blue, red edge, and near infrared. In multispectral imaging, each spectral channel corresponds to a specific spectral range, which is used to capture the spectral channel images within the range, so as to form a multispectral image in subsequent processing. Since there are slight differences in the capture times of different spectral channels, the disorder of the sampling points increases;

[0087] The values ​​of the preset environmental impact coefficient and the preset synchronization coefficient can be determined by the user according to the actual application scenario. The larger the value of the preset environmental impact coefficient and the smaller the value of the preset synchronization coefficient, the greater the user's need to determine the sampling point disorder coefficient according to the sampling point radiation value. Provide a value of the preset environmental impact coefficient and the preset synchronization coefficient, detect the historical records of determining the sampling point disorder coefficient according to the sampling point radiation value, record the average value of the environmental impact coefficient corresponding to the historical records that can meet the user's needs as the preset environmental impact coefficient, and record the average value of the synchronization coefficient corresponding to the historical records that can meet the user's needs as the preset synchronization coefficient;

[0088] The characteristic line turbulence amplitude is the area of ​​the smallest rectangle that can contain each characteristic line, and the disorder threshold = sampling point disorder coefficient + characteristic line turbulence amplitude;

[0089] The values ​​of the preset sampling point disorder coefficient and the preset characteristic line turbulence amplitude can be determined by the user according to the actual application scenario. The smaller the values ​​of the preset sampling point disorder coefficient and the preset characteristic line turbulence amplitude are, the greater the user's demand for uniform division according to the disorder threshold is. A value of the preset sampling point disorder coefficient and the preset characteristic line turbulence amplitude is provided, and the historical records of the user's uniform division according to the disorder threshold are detected. The average value of the sampling point disorder coefficient corresponding to the historical records that can meet the user's needs is recorded as the preset sampling point disorder coefficient, and the average value of the characteristic line turbulence amplitude corresponding to the historical records that can meet the user's needs is recorded as the preset characteristic line turbulence amplitude;

[0090] When the image division method is uniform division according to the confusion threshold, the first spectral image is divided into a number of rectangular areas with the same area and shape, each rectangular area is recorded as a division unit, and the number of division units is positively correlated with the confusion threshold;

[0091] When the image division method is to divide according to the position of the characteristic line, the first spectral image is divided into a first division unit and a second division unit, the first division unit is the smallest rectangle that can contain each characteristic line, and the second division unit is other areas outside the first division unit in the first spectral image.

[0092] Specifically, the sampling point window area is determined according to the sampling point mutation coefficient of the division unit, where:

[0093] The sampling point window area corresponding to a single sampling point is negatively correlated with the sampling point mutation coefficient corresponding to the division unit where the sampling point is located;

[0094] The sampling point window corresponding to a single sampling point is a square area centered on the sampling point.

[0095] Among them, the sampling point mutation coefficient corresponding to a single division unit = the number of mutation sampling points in the division unit + the average value of the color intensity values ​​corresponding to each mutation sampling point in the division unit;

[0096] It should be noted that the four sides of the sampling point window corresponding to each sampling point are vertical or parallel to the four borders corresponding to the first spectral image. It can be understood that the first spectral image is a rectangle, and the borders of the first spectral image are the four sides of the rectangle.

[0097] Specifically, a window analysis is performed on each sampling point in the first spectral image, wherein:

[0098] When performing window analysis on a single sampling point, the sampling point is recorded as the target sampling point;

[0099] If the window influence coefficient is greater than or equal to the preset window influence coefficient and the target sampling point satisfies the preset window state, then the target sampling point is a special point;

[0100] The preset window condition is that the color intensity value of the target sampling point is the maximum value among the color intensity values ​​corresponding to each sampling point in the sampling point window corresponding to the target sampling point.

[0101] Wherein, whether the target sampling point is a special point is determined according to the window influence coefficient corresponding to the target sampling point and whether the target sampling point meets the preset window condition. If the window influence coefficient is less than the preset window influence coefficient or the target sampling point does not meet the preset window state, the target sampling point is not a special point. The value of the preset window influence coefficient can be determined by the user according to the actual application scenario. The greater the user's demand for improving the recognition accuracy of special points, the smaller the value of the preset window influence coefficient is. A value of the preset window influence coefficient is provided, and the historical records of judging the sampling point as a special point are detected, and the average value of the window influence coefficient corresponding to the historical records that can meet the user's needs is recorded as the preset window influence coefficient.

[0102] It should be noted that if there is no sampling point window for the sampling point, the sampling points whose angle disorder value of the same sampling point is less than the preset angle disorder value of the same sampling point and whose proportion of the same sampling points is within the preset proportion range of the same sampling points are recorded as special points.

[0103] Specifically, the window influence coefficient corresponding to a single sampling point is determined according to whether the characteristic window line corresponding to the sampling point meets the preset placement conditions and the proportion of the same sampling points, where:

[0104] If the preset placement conditions are met or the proportion of the same sampling points is not within the preset range of the same sampling point proportion, the window influence coefficient is determined according to the specific index;

[0105] If the preset placement conditions are not met and the proportion of the same sampling points is within the preset range of the proportion of the same sampling points, the window influence coefficient is determined according to the angle disorder value of the same sampling points;

[0106] When determining the window influence coefficient according to the same sampling point angle disorder value, if the same sampling point angle disorder value is greater than or equal to the preset same sampling point angle disorder value, the window influence coefficient is determined according to the same sampling point angle disorder value; if the same sampling point angle disorder value is less than the preset same sampling point angle disorder value, the window influence coefficient is determined according to the regional disorder degree and the local similarity;

[0107] The preset placement condition is that the line segment threshold is a standard threshold and the window line position relationship is a cross relationship.

[0108] Wherein, for a single sampling point, the sampling point is recorded as a first target sampling point, and four line segments of a cross line passing through the first target sampling point in a sampling point window corresponding to the first target sampling point are recorded as characteristic window lines;

[0109] The line segment threshold is confirmed in the following way: the brightness mean corresponding to each feature window line corresponding to the first target sampling point is detected, and the set of feature window lines whose brightness mean difference is less than the preset brightness difference is recorded as the same line segment group. The line segment threshold is the maximum value of the number of feature window lines corresponding to each same line segment group, and the standard threshold is 2; the window line position relationship is a cross relationship, that is, the angle between the two feature window lines is 90°; the brightness mean is the average value of the color intensity values ​​corresponding to each sampling point in a single feature window line. For any two feature window lines, the larger value of the brightness mean corresponding to the two feature window lines is recorded as m1, and the smaller value is recorded as m2, and the brightness mean difference = (m1-m2) / m1;

[0110] The sampling points adjacent to the first target sampling point are recorded as first reference sampling points, and the first reference sampling points with the same color intensity value as the first target sampling point are recorded as identical sampling points, and the identical sampling point ratio = the number of identical sampling points / the total number of first reference sampling points; the values ​​within the preset identical sampling point ratio range are all greater than or equal to the first preset identical sampling point ratio and less than the second preset identical sampling point ratio, wherein the first preset identical sampling point ratio is less than the second preset identical sampling point ratio;

[0111] When the preset placement condition is not met and the proportion of the same sampling points is within the preset proportion range of the same sampling points, the window influence coefficient is determined according to the angle disorder value of the same sampling points, wherein, if the angle disorder value of the same sampling points is greater than or equal to the preset angle disorder value of the same sampling points, the window influence coefficient is negatively correlated with the angle disorder value of the same sampling points, and if the angle disorder value of the same sampling points is less than the preset angle disorder value of the same sampling points, the window influence coefficient = regional disorder + local similarity;

[0112] When the preset placement conditions are met or the proportion of the same sampling points is not within the preset range of the same sampling point proportion, the window influence coefficient is positively correlated with the specific index;

[0113] The method for confirming the angle disorder value of the same sampling point is that, for any two sampling point directions, the minimum angle between the two sampling point directions is recorded as the reference angle, the angle of each reference angle is identified, and the reference angle with the largest angle is recorded as the influence angle; the angle disorder value of the same sampling point = (the total number of the first reference sampling points within the influence angle - the number of the same sampling points within the influence angle) / the total number of the first reference sampling points within the influence angle; the method for confirming the direction of the sampling point is that, for a single first reference sampling point, the direction of the ray that takes the position of the first target sampling point as the endpoint and passes through the position of the first reference sampling point is the sampling point direction;

[0114] The regional disorder degree and the local similarity are confirmed in the following manner: the region including the same sampling points in the closed region formed by the sampling point window corresponding to the influencing angle and the first target sampling point is recorded as the first region, and the other regions outside the first region in the sampling point window corresponding to the first target sampling point are recorded as the second region, the regional disorder degree = |the average value of the color intensity values ​​corresponding to the built-in sampling points in the first region - the average value of the color intensity values ​​corresponding to the built-in sampling points in the second region|, the built-in sampling points are the other sampling points in the sampling point window except the sampling points located on the ray corresponding to the influencing angle, and the local similarity is the fluctuation value of the color intensity values ​​corresponding to the built-in sampling points in the first region;

[0115] Specificity index = correlation coefficient - proportion of identical sampling points. The correlation coefficient is confirmed by recording a sampling point as a target sampling point, recording other sampling points in the sampling point window where the target sampling point is located as reference sampling points, recording the reference sampling points in the same row as the target sampling point as row sampling points, and recording the reference sampling points in the same column as the target sampling point as column sampling points. Correlation coefficient = |column fluctuation threshold - row fluctuation threshold|, where the column fluctuation threshold is the standard deviation of the color intensity values ​​corresponding to the sampling points in each column, and the row fluctuation threshold is the standard deviation of the color intensity values ​​corresponding to the sampling points in each row.

[0116] The values ​​of the preset same sampling point angle disorder value, the preset brightness mean difference, the first preset same sampling point ratio and the second preset same sampling point ratio can be determined by the user according to the actual application scenario. The smaller the value of the preset same sampling point angle disorder value is, the greater the user's demand for determining the window influence coefficient according to the same sampling point angle disorder value is. A value of the preset same sampling point angle disorder value is provided, and the historical records of determining the window influence coefficient according to the same sampling point angle disorder value are detected, and the same sampling point angle disorder value corresponding to the historical records that can meet the user's needs is recorded as the preset same sampling point angle disorder value. The sample point angle disorder value, the greater the user's demand for the similarity of the cross lines in the same line segment group, the smaller the value of the preset brightness mean difference is, and a preset brightness mean difference is provided. The preset brightness mean difference is 30%. The greater the user's demand for determining the window influence coefficient according to the specific index, the greater the value of the first preset proportion of the same sampling points, and the smaller the value of the second preset proportion of the same sampling points. A first preset proportion of the same sampling points and a second preset proportion of the same sampling points are provided. The first preset proportion of the same sampling points is 40%, and the second preset proportion of the same sampling points is 70%.

[0117] Specifically, whether the special point cluster area is a building area is determined based on the brightness difference coefficient and the number of special points, where:

[0118] If the brightness difference coefficient is less than the preset brightness difference coefficient and the number of special points is greater than or equal to the preset number of special points, the special point clustering area is a building area;

[0119] The special point clustering area is determined according to the special point threshold similarity coefficient and the special point spacing.

[0120] Among them, the brightness difference coefficient is the standard deviation of the color intensity values ​​corresponding to each sampling point in a single special point cluster area; the number of special points is the total number of special points in a single special point cluster area;

[0121] The values ​​of the preset brightness difference coefficient and the preset number of special points can be determined by the user according to the actual application scenario. The greater the user's demand for the accuracy of determining the building area, the smaller the value of the preset brightness difference coefficient and the larger the value of the preset number of special points. Provide a value of the preset brightness difference coefficient and the preset number of special points, detect the historical records of determining the special point cluster area as the building area, record the average value of the brightness difference coefficient corresponding to the historical records that can meet the user's needs as the preset brightness difference coefficient, and record the average value of the number of special points corresponding to the historical records that can meet the user's needs as the preset number of special points;

[0122] The special point clustering area is determined according to the special point threshold similarity coefficient and the special point spacing, wherein cluster analysis is performed on each special point. When cluster analysis is performed on a single special point, the special point is recorded as a target special point, and other special points except the target special point are recorded as reference special points. The set of reference special points and target special points whose special point threshold similarity coefficient with the target special point is greater than the preset special point threshold similarity coefficient and whose special point spacing is less than the preset special point spacing is recorded as a special point set, and the minimum rectangle that can contain each special point in the single special point set is recorded as a special point clustering area, and cluster analysis is continued for the special points that are not recorded in the special point clustering area until all special points are recorded in the special point clustering area, and the cluster analysis is stopped;

[0123] For two special points, the larger value of the color intensity values ​​corresponding to the two special points is recorded as a1, and the smaller value is recorded as a2. The special point threshold similarity coefficient = 1 / (a1-a2), and the special point spacing is the shortest distance from one special point to another special point;

[0124] The values ​​of the preset special point threshold similarity coefficient and the preset special point spacing can be determined by the user according to the actual application scenario. The greater the user's demand for improving the similarity of special points in the special point clustering area, the greater the value of the preset special point threshold similarity coefficient, and the smaller the value of the preset special point spacing. A value of the preset special point threshold similarity coefficient and the preset special point spacing is provided, and the historical records of determining the special point clustering area according to the special point threshold similarity coefficient and the special point spacing are detected, and the average value of the special point threshold similarity coefficient corresponding to the historical records that can meet the user's needs is recorded as the preset special point threshold similarity coefficient, and the average value of the special point spacing corresponding to the historical records that can meet the user's needs is recorded as the preset special point spacing.

[0125] Specifically, the analysis method is determined according to the image difference and the building area assessment coefficient, where:

[0126] If the image difference is greater than or equal to the preset image difference or the building area assessment coefficient is greater than or equal to the preset building area assessment coefficient, the analysis method is to determine the stability determination method according to the comparison area difference and the difference area mapping coefficient;

[0127] If the image difference is less than the preset image difference and the building area assessment coefficient is less than the preset building area assessment coefficient, the analysis method is to determine the stability according to the damage threshold of the building area;

[0128] The stability of a single building area is negatively correlated with the damage threshold of the building area.

[0129] The image difference is confirmed by recording the larger value of the image thresholds corresponding to the second spectral image and the first spectral image as k1 and the smaller value as k2. Image difference = (k1-k2) / k1. For a single spectral image, the image threshold corresponding to the spectral image = the number of mutation sampling points in the spectral image + the standard deviation of the color intensity values ​​corresponding to each mutation sampling point in the spectral image.

[0130] Building area assessment coefficient = number of building areas in the first spectral image + building area distribution coefficient in the first spectral image, average value of distance reference values ​​corresponding to each building area, for a single building area, record the building area as the target area, record other building areas outside the target area as reference areas, record the minimum value of the shortest distance from the center position of the target area to the center position of each reference area as the distance reference value, and the center position of a single building area is the center of the circumscribed circle of the building area;

[0131] The values ​​of the preset image difference and the preset building area evaluation coefficient can be determined by the user according to the actual application scenario. The greater the user's demand for determining the stability determination method according to the comparison area difference and the difference area mapping coefficient, the smaller the values ​​of the preset image difference and the preset building area evaluation coefficient are. A value of the preset image difference and the preset building area evaluation coefficient is provided, the preset image difference is 60%, and the historical records of determining the stability determination method according to the comparison area difference and the difference area mapping coefficient are detected, and the average value of the building area evaluation coefficient corresponding to the historical records that can meet the user's needs is recorded as the preset building area evaluation coefficient;

[0132] The damage threshold is the average value of the reference threshold values ​​corresponding to each damage area in a single building area. For a single damage area, the damage area is recorded as the target damage area, and other damage areas outside the target damage area are recorded as reference damage areas. The average value of the shortest distance from the center position corresponding to the target damage area to the center position corresponding to each reference damage area is recorded as the reference threshold.

[0133] The method for confirming the damaged area is to perform a comparison analysis on each difference sampling point in a single matching area. When performing the comparison analysis on a single difference sampling point, the difference sampling point is recorded as a target difference sampling point, and other difference sampling points except the target difference sampling point are recorded as reference difference sampling points. The set of reference difference sampling points and target difference sampling points whose difference sampling point spacing with the target difference sampling point is less than the preset difference sampling point spacing is recorded as a difference sampling point set. The minimum rectangle that can contain each difference sampling point in the single difference sampling point set is recorded as a damaged area, and the comparison analysis is continued for the difference sampling points that are not recorded in the damaged area until all the difference sampling points are recorded in the damaged area, and then the comparison analysis is stopped;

[0134] The difference sampling point spacing is the shortest distance from one difference sampling point to another difference sampling point. The value of the preset difference sampling point spacing can be determined by the user according to the actual application scenario. The greater the user's demand for improving the accuracy of damaged area recognition, the smaller the value of the preset difference sampling point spacing. A value of the preset difference sampling point spacing is provided, and the historical records of determining the damaged area based on the difference sampling point spacing are detected. The average value of the difference sampling point spacing corresponding to the historical records that can meet the user's needs is recorded as the preset difference sampling point spacing.

[0135] Specifically, the stability determination method is determined according to the comparison area difference and the difference area mapping coefficient, wherein:

[0136] If the comparison area difference is greater than or equal to the preset comparison area difference or the difference area mapping coefficient is greater than or equal to the preset difference area mapping coefficient, the stability is determined according to the comparison area difference;

[0137] If the comparison region difference is less than the preset comparison region difference and the difference region mapping coefficient is less than the preset difference region mapping coefficient, the stability is determined according to the neighboring region influence coefficient and the comparison region difference.

[0138] Among them, the first spectral image and the second spectral image are aligned to the same coordinate system through image registration technology. This is easy for those skilled in the art to understand and will not be described in detail. The area in the second spectral image that overlaps with the building area in the first spectral image is recorded as the matching area. The method for confirming the difference degree of the comparison area is that, for a single building area in the first spectral image and its corresponding matching area, the difference degree of the comparison area = the number of difference sampling points in the matching area / the total number of sampling points in the building area. For a sampling point in the matching area, the sampling point is recorded as the first sampling point, and the sampling point in the building area that is located at the same position as the sampling point to be analyzed is recorded as the second sampling point. If the color intensity values ​​of the first sampling point and the second sampling point are greater than the preset difference, the first sampling point is the difference sampling point;

[0139] The difference building area is a building area whose comparison area difference in the first spectral image is greater than the preset comparison area difference. The difference area mapping coefficient is the average value of the difference distances corresponding to each difference area. For a single difference building area, the difference building area is recorded as a target difference area, and other difference building areas outside the target difference area are recorded as reference difference areas. The minimum value of the shortest distances from the center position of the target difference area to the center position of each reference difference area is recorded as the difference distance.

[0140] The values ​​of the preset comparison area difference and the preset difference area mapping coefficient can be determined by the user according to the actual application scenario. The smaller the values ​​of the preset comparison area difference and the preset difference area mapping coefficient are, the greater the user's demand for determining the stability according to the comparison area difference is. A value of the preset comparison area difference and the preset difference area mapping coefficient is provided. The preset comparison area difference is 60%. The historical records of determining the stability according to the comparison area difference are detected, and the average value of the difference area mapping coefficient corresponding to the historical records that can meet the user's needs is recorded as the preset difference area mapping coefficient;

[0141] If the comparison area difference is greater than or equal to the preset comparison area difference or the difference area mapping coefficient is greater than or equal to the preset difference area mapping coefficient, the stability corresponding to the single building area is negatively correlated with the comparison area difference corresponding to the building area;

[0142] If the comparison area difference is less than the preset comparison area difference and the difference area mapping coefficient is less than the preset difference area mapping coefficient, the stability corresponding to the single building area = 1 / (the comparison area difference corresponding to the building area + the influence coefficient of the adjacent area);

[0143] The method for confirming the neighboring area influence coefficient is that, for a single building area in the first spectral image, the building area is recorded as the target building area, and other building areas outside the target building area in the first spectral image are recorded as reference building areas. The neighboring area influence coefficient is the average value of the sub-influence coefficients corresponding to each reference building area, and the sub-influence coefficient corresponding to a single reference building area = the corresponding comparison area difference of the reference building area / the shortest distance from the center position of the reference building area to the center position of the target building area.

[0144] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

[0145] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A building structure stability detection method based on multispectral images, characterized in that: include: Acquire spectral image information, determine the image state according to the background coefficient and the building density coefficient, and determine the processing method according to the image state, the processing method is to determine the building area according to the special point clustering area, or to determine the building area according to the characteristic line; When determining the building area according to the special point cluster area, determine the image division method according to the sampling point disorder coefficient and the characteristic line turbulence amplitude to obtain a number of division units, and determine the sampling point window area according to the sampling point mutation coefficient of the division unit, use the determined sampling point window area to perform window analysis on each sampling point in the first spectral image to determine the special point, and determine whether the special point cluster area is a building area according to the brightness difference coefficient and the number of special points; When determining the building area based on feature lines, the building area is determined based on the closed area formed by the feature lines; When the building area determination is completed, an analysis method is determined according to the image difference between the second spectral image and the first spectral image and the building area assessment coefficient of the first spectral image. The analysis method is to determine the stability determination method according to the comparison area difference and the difference area mapping coefficient or to determine the stability according to the damage threshold of the building area; The stability determination method is to determine the stability according to the difference of the comparison area or to determine the stability according to the influence coefficient of the adjacent area and the difference of the comparison area; The image division method is to divide the image evenly according to the disorder threshold or to divide the image according to the position of the feature line.

2. The building structure stability detection method based on multispectral images according to claim 1 is characterized in that: The image state is determined based on the background coefficient and the building density coefficient. The image state includes: A first image state in which a background coefficient is greater than or equal to a preset background coefficient or a building closeness coefficient is greater than or equal to a preset building closeness coefficient; A second image state in which the background coefficient is less than a preset background coefficient and the building density coefficient is less than a preset building density coefficient.

3. The building structure stability detection method based on multispectral images according to claim 2 is characterized in that: Determine the processing method according to the image status, where: In the first image state, the processing method is to determine the building area based on the special point cluster area; In the second image state, the processing method is to determine the building area according to the feature lines.

4. The building structure stability detection method based on multispectral images according to claim 3 is characterized in that: The image division method is determined according to the sampling point disorder coefficient and the characteristic line turbulence amplitude to obtain a number of division units, among which, If the sampling point disorder coefficient is greater than or equal to the preset sampling point disorder coefficient or the characteristic line turbulence amplitude is greater than or equal to the preset characteristic line turbulence amplitude, the image division method is to divide the image uniformly according to the disorder threshold; If the sampling point disorder coefficient is less than the preset sampling point disorder coefficient and the characteristic line turbulence amplitude is less than the preset characteristic line turbulence amplitude, the image division method is to divide according to the characteristic line position.

5. The building structure stability detection method based on multispectral images according to claim 4 is characterized in that: The sampling point window area is determined according to the sampling point mutation coefficient of the division unit, where: The sampling point window area corresponding to a single sampling point is negatively correlated with the sampling point mutation coefficient corresponding to the division unit where the sampling point is located; The sampling point window corresponding to a single sampling point is a square area centered on the sampling point.

6. The building structure stability detection method based on multispectral images according to claim 5 is characterized in that: A window analysis is performed on each sampling point in the first spectral image, wherein: When performing window analysis on a single sampling point, the sampling point is recorded as the target sampling point; If the window influence coefficient is greater than or equal to the preset window influence coefficient and the target sampling point satisfies the preset window state, then the target sampling point is a special point; The preset window condition is that the color intensity value of the target sampling point is the maximum value among the color intensity values ​​corresponding to each sampling point in the sampling point window corresponding to the target sampling point.

7. The building structure stability detection method based on multispectral images according to claim 6 is characterized in that: The window influence coefficient corresponding to a single sampling point is determined based on whether the characteristic window line corresponding to the sampling point meets the preset placement conditions and the proportion of the same sampling points, where: If the preset placement conditions are met or the proportion of the same sampling points is not within the preset range of the same sampling point proportion, the window influence coefficient is determined according to the specific index; If the preset placement conditions are not met and the proportion of the same sampling points is within the preset range of the proportion of the same sampling points, the window influence coefficient is determined according to the angle disorder value of the same sampling points; When determining the window influence coefficient according to the same sampling point angle disorder value, if the same sampling point angle disorder value is greater than or equal to the preset same sampling point angle disorder value, the window influence coefficient is determined according to the same sampling point angle disorder value; if the same sampling point angle disorder value is less than the preset same sampling point angle disorder value, the window influence coefficient is determined according to the regional disorder degree and the local similarity; The preset placement condition is that the line segment threshold is a standard threshold and the window line position relationship is a cross relationship.

8. The building structure stability detection method based on multispectral images according to claim 6 is characterized in that: Whether the special point cluster area is a building area is determined based on the brightness difference coefficient and the number of special points, where: If the brightness difference coefficient is less than the preset brightness difference coefficient and the number of special points is greater than or equal to the preset number of special points, the special point clustering area is a building area; The special point clustering area is determined according to the special point threshold similarity coefficient and the special point spacing.

9. The building structure stability detection method based on multispectral images according to claim 8 is characterized in that: The analysis method is determined based on the image difference and the building area assessment coefficient, where: If the image difference is greater than or equal to the preset image difference or the building area assessment coefficient is greater than or equal to the preset building area assessment coefficient, the analysis method is to determine the stability determination method according to the comparison area difference and the difference area mapping coefficient; If the image difference is less than the preset image difference and the building area assessment coefficient is less than the preset building area assessment coefficient, the analysis method is to determine the stability according to the damage threshold of the building area; The stability of a single building area is negatively correlated with the damage threshold of the building area.

10. The building structure stability detection method based on multispectral images according to claim 9 is characterized in that: The stability determination method is determined according to the comparison area difference and the difference area mapping coefficient, where: If the comparison area difference is greater than or equal to the preset comparison area difference or the difference area mapping coefficient is greater than or equal to the preset difference area mapping coefficient, the stability is determined according to the comparison area difference; If the comparison region difference is less than the preset comparison region difference and the difference region mapping coefficient is less than the preset difference region mapping coefficient, the stability is determined according to the neighboring region influence coefficient and the comparison region difference.

Images