Remote sensing image shadow index performance evaluation method based on multiple scenes and multiple criteria

Abstract

The application discloses a remote sensing image shadow index performance evaluation method based on multiple scenes and multiple criteria, so as to evaluate the performance of different shadow indexes in urban and mountain scenes. The method comprises the following steps: data preparation, shadow extraction, sample selection, multiple criteria evaluation and comprehensive scoring. In the multiple criteria evaluation process, the shadow extraction accuracy evaluation is for the urban and mountain scenes; the separability of the shadow and the easily confused ground objects is for the urban scene, and is quantitatively evaluated by a 1.5 interquartile range (1.5IQR) separation ratio; the sensitivity of the shadow to the light intensity is for the mountain scene, and is quantitatively evaluated by the correlation between the shadow index and the cosine value of the solar incident angle (cosi). The method disclosed by the application can avoid the problems of single scene and single criterion in the traditional evaluation method, the evaluation result is intuitive and reliable, and the method has important scientific significance for objectively analyzing and evaluating the performance of the shadow index.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of remote sensing image index performance evaluation, and particularly relates to a remote sensing image shadow index performance evaluation method based on multiple scenes and multiple criteria. BACKGROUND

[0002] Shadows are ubiquitous in remote sensing images, and with the increase of image spatial resolution, shadows become more and more obvious, especially in high latitude areas or images taken in winter when the sun's elevation angle is low. Shadows not only affect the accuracy of image target recognition, image matching and change detection, but also affect the retrieval of vegetation information in remote sensing. Therefore, shadow detection is a necessary step before applying these tasks. For medium and high resolution images in mountainous areas, shadows are mainly composed of terrain shadows, including areas that are not illuminated by sunlight or are blocked by other mountains. In contrast, shadows in urban areas are usually caused by buildings or trees casting shadows on other ground objects. Due to the complexity of land cover types in urban areas, the shadows cast on the ground objects are very similar to the spectral characteristics of the ground objects themselves, which easily leads to a decrease in the accuracy of shadow detection.

[0003] Among the methods of shadow detection, the index method is a simple and effective method. So far, many shadow indices have been widely used. However, due to the different construction methods of various shadow indices and the different color spaces they rely on, there are significant differences in the results of shadow extraction by different shadow indices. In addition, due to the different causes of shadows in urban and mountainous areas, there is considerable uncertainty in the shadow detection effect of different shadow indices in different scenes.

[0004] In summary, providing a remote sensing image shadow index performance evaluation method based on multiple scenes and multiple criteria has important scientific significance and practical value for objectively evaluating the performance of each shadow index in different scenes. SUMMARY

[0005] The technical problem to be solved by the present application is to overcome the shortcomings of the prior art and provide a remote sensing image shadow index performance evaluation method based on multiple scenes and multiple criteria. To solve the above technical problems, the basic idea of the technical solution adopted by the present application is:

[0006] The remote sensing image shadow index performance evaluation method based on multiple scenes and multiple criteria comprises the following steps:

[0007] Step 1, data preparation and binary classification: obtain multispectral image data of urban scenes and mountain scenes, and digital elevation model (DEM) data corresponding to the mountain scenes, and use the shadow index to be evaluated to extract shadows in the images;

[0008] Step 2, sample selection: select appropriate shadow samples and non-shadow samples in the image, wherein the non-shadow samples in the urban scene are subdivided into 6 types of easily confused ground objects, including water, vegetation, soil, light-colored impervious surface, dark-colored impervious surface and blue-colored impervious surface, and the non-shadow samples in the mountainous scene are not subdivided;

[0009] Step 3, separability evaluation of shadow and easily confused ground objects: according to the shadow samples and easily confused ground object samples in the mountainous scene, the 1.5IQR separation ratio of shadow and each type of easily confused ground object is calculated, and a box plot is drawn to show the separation of shadow and each type of easily confused ground object, and finally the average 1.5IQR separation ratio of the 6 types of easily confused ground objects is taken as the 1.5IQR separation ratio of shadow and easily confused ground objects;

[0010] Step 4, sensitivity evaluation of shadow index to light intensity: the determination coefficient R 2 of the shadow index value and the cosine value of the solar incident angle (cosi) of all shadow samples and non-shadow samples in the mountainous scene is calculated, and a density scatter plot is drawn to show the change and distribution of the shadow index value with the light intensity;

[0011] Step 5, accuracy evaluation of shadow extraction: according to all shadow samples and non-shadow samples in the urban and mountainous scenes, the accuracy R a of the shadow index to shadow extraction in each scene is calculated respectively.

[0012] Step 6, comprehensive scoring: the quantitative calculation results in the multi-criteria evaluation are summarized and scored, and the scoring results are shown by drawing a heat map to show the performance of the shadow index in each criterion.

[0013] Further, in step 3, the 1.5IQR separation ratio is calculated according to the following formula:

[0014]

[0015] 1.5IQRSR=1-1.5IQRCR

[0016] In the formula, 1.5IQRCR is the 1.5IQR confusion ratio, 1.5IQRSR is the 1.5IQR separation ratio, Q 3_c is the upper quartile of the easily confused ground object (Q3), Q 1_s is the lower quartile of the shadow (Q1), and 1.5IQR _c is the 1.5IQR range of the easily confused ground object.

[0017] Further, in step 5, the accuracy R a of the shadow index to shadow extraction is calculated according to the following formula:

[0018]

[0019] In the formula, R a is the accuracy of shadow extraction, N tds is the number of real shadow pixels in the detected shadow, N tdns is the number of real non-shadow pixels in the detected non-shadow, N total is the total number of pixels of the sample.

[0020] Further, in step 6, the comprehensive scoring items include: the 1.5IQR separation ratio of the shadow and the easily confused ground objects in the urban scene, the determination coefficient R 2 of the shadow index value and the cosine value (cosi) of the sun incident angle in the mountainous scene, the shadow extraction accuracy R a in the urban scene and the mountainous scene. The comprehensive score is the average value of all items, and the higher the score value, the better the performance of the index.

[0021] After the above technical solution, the present application has the following beneficial effects compared with the prior art.

[0022] The multi-scene and multi-criteria based remote sensing image shadow index performance evaluation method provided by the present application introduces the cosine (cosi) of the sun incident angle to evaluate the sensitivity of the index to the illumination intensity in addition to the conventional shadow extraction accuracy evaluation in the multi-criteria evaluation process, and evaluates the separability of the shadow and the easily confused ground objects by calculating the 1.5IQR separation ratio of the shadow and the easily confused ground objects. The multi-scale evaluation includes the mountainous scene and the urban scene, the sensitivity to the illumination intensity is used for evaluating the mountainous scene, and the ability to separate the easily confused ground objects is used for evaluating the urban scene. Finally, the results obtained from each evaluation criterion are summarized to form a multi-criteria evaluation score of the shadow index, so as to clearly and objectively show the performance of each shadow index in different scenes and different criteria.

[0023] The specific embodiments of the present application will be further described in detail below with reference to the accompanying drawings. BRIEF DESCRIPTION OF DRAWINGS

[0024] The accompanying drawings, which are part of the present application, serve to provide a further understanding of the present application, and the schematic embodiments of the present application and the descriptions thereof serve to explain the present application, but do not constitute an improper limitation on the present application. Obviously, the accompanying drawings in the following description are only some embodiments, and other drawings can be obtained by those skilled in the art without creative labor. In the drawings:

[0025] Figure 1 is a flowchart of the present application;

[0026] Figure 2 is a schematic diagram of multi-spectral image data of an embodiment of the present application;

[0027] Figure 3 is a binary image of shadow extraction result of the embodiment of the present application;

[0028] Figure 4 is a box plot of index value of shadow and easily confused ground objects (Shadow represents shadow, Water represents water body, Soil represents soil, Vegetation represents vegetation, BIS represents blue impervious surface, DIS represents dark impervious surface, and LIS represents light impervious surface) of the embodiment of the present application;

[0029] Figure 5 is a density scatter plot of index value of shadow and cosine value of solar incident angle (cosi) of the embodiment of the present application and its determination coefficient (R 2 );

[0030] Figure 6 is a heat map of comprehensive score result of the embodiment of the present application.

[0031] It should be noted that these drawings and written descriptions are not intended to limit the scope of the concept of the present application in any way, but to illustrate the concept of the present application to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION

[0032] In order to make the purpose, technical scheme and advantages of the embodiments of the present application more clear, the technical scheme in the embodiments will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. The following embodiments are used to illustrate the present application, but not to limit the scope of the present application.

[0033] Embodiment one

[0034] As shown in Figure 1 , the shadow index performance evaluation method for remote sensing image based on multiple scenes and multiple criteria comprises the following steps:

[0035] Step 1, obtain multispectral image data of urban scene and mountain scene, and digital elevation model (DEM) data corresponding to the mountain scene, and use the shadow index to be evaluated to extract the shadow in the image, as shown in Figure 2 and Figure 3 .

[0036] Specifically, Landsat8 OLI multispectral images of Gaoligong Mountain area in Yunnan Province and Yinchuan City in Ningxia Hui Autonomous Region are obtained, and the imaging times are January 22, 2014 and November 21, 2013, respectively. Figure 2 (a) is the image of mountain area (Gaoligong Mountain), which is used to evaluate the performance of shadow index in mountain scene; Figure 2(b) is the urban area (Yinchuan City) image, used to evaluate the performance of shadow index in urban scenes. Both images are radiometrically calibrated and dark pixel method atmospheric correction processed.

[0037] Take the combined shadow index (CSI), normalized saturation- value difference index (NSVDI), normalized dark pixel index (NDPI) and shadow detection index (SDI) as examples, shadow extraction is performed on remote sensing images of urban scenes and mountain scenes, as shown in Figure 3 Figure 3 (a)- Figure 3 (d) are the shadow extraction results of CSI, NSVDI, NDPI and SDI index in urban scenes, respectively; Figure 3 (e)- Figure 3 (h) are the shadow extraction results of CSI, NSVDI, NDPI and SDI index in mountain scenes, respectively. In addition, Figure 3 the black part in (f) represents shadow, and the white part represents non-shadow.

[0038] Step 2, select appropriate shadow samples and non-shadow samples in the image, wherein the non-shadow samples in urban scenes are subdivided into 6 types of easily confused ground objects, including water, vegetation, soil, light-colored impervious surface, dark-colored impervious surface and blue-colored impervious surface, and the non-shadow samples in mountain scenes are not subdivided.

[0039] Step 3, separability evaluation of shadow and easily confused ground objects: according to the shadow samples and easily confused ground object samples in mountain scenes, the 1.5 times interquartile range (1.5IQR) separation ratio of shadow and each type of easily confused ground object is calculated, and a box plot is drawn to show the separation of shadow and each type of easily confused ground object. Finally, the average 1.5IQR separation ratio of the 6 types of easily confused ground objects is taken as the 1.5IQR separation ratio of shadow and easily confused ground objects.

[0040] Specifically, the 1.5IQR separation ratio is calculated according to the following formula:

[0041]

[0042] 1.5IQRSR=1-1.5IQRCR

[0043] In the formula, 1.5IQRCR is the 1.5IQR confusion ratio, 1.5IQRSR is the 1.5IQR separation ratio, Q 3_c is the upper quartile of the easily confused ground object (Q3), Q 1_s is the lower quartile of the shadow (Q1), and 1.5IQR _c is the 1.5IQR range of the easily confused ground object.

[0044] Figure 4 (a)- Figure 4 ​(d) Boxplot of the index value distribution of CSI, NSVDI, NDPI and SDI indices for the 6 categories of confusing ground objects described in Step 3.

[0045] Table 1. 1.5IQR separation ratio of shadow and confusing ground objects

[0046]

[0047] Table 1. 1.5IQR separation ratio of shadow and confusing ground objects

[0048] Step 4, Sensitivity evaluation of shadow indices to illumination intensity: Calculate the determination coefficient R of shadow indices value and cosine of solar incident angle (cosi) of all shadow samples and non-shadow samples in mountainous scene 2 , and plot the density scatter plot to show the variation and distribution of shadow indices value with illumination intensity.

[0049] Figure 5 (a)- Figure 5 (d) Density scatter plot and determination coefficient R between the index value of CSI, NSVDI, NDPI and SDI indices and cosine of solar incident angle (cosi) 2 , R 2 is larger, the sensitivity of shadow indices to illumination intensity is higher.

[0050] Step 5, Calculate the accuracy rate R of shadow indices to shadow extraction in each scene according to all shadow samples and non-shadow samples in urban and mountainous scenes a , as shown in Table 2.

[0051] Table 2. Accuracy rate of shadow extraction

[0052]

[0053] Specifically, the accuracy rate R of shadow indices to shadow extraction a is calculated according to the following formula:

[0054]

[0055] In the formula, R a is the accuracy rate of shadow extraction, N tds is the number of real shadow pixels in detected shadow, N tdns is the number of real non-shadow pixels in detected non-shadow, N total is the total number of pixels of samples, and R a value is larger, the accuracy rate of the index to shadow extraction is higher.

[0056] Step 6, the quantitative calculation results in the multi-criteria evaluation are summarized and comprehensive scoring is performed, and the scoring results are shown by drawing a heat map to show the performance of the shadow index in each criterion, as shown in Figure 6 .

[0057] Specifically, the comprehensive scoring items include: the 1.5IQR separation ratio of the shadow and the easily confused ground objects in the urban scene in the step 3 (the average value in Table 1); the determination coefficient R 2 of the shadow index value and the cosine value of the solar incident angle (cosi) in the mountain scene in the step 4; and the shadow extraction accuracy R a in the urban scene and the mountain scene in the step 5. Figure 6 The comprehensive score is the average value of all items, and the higher the score value is, the better the comprehensive performance of the index is.

[0058] According to the multi-criteria evaluation results of Figure 6 , the performance of each shadow index in different scenes and different criteria can be clearly evaluated.

[0059] In summary, the present application can avoid the problems of single scene and single criterion in the traditional evaluation method, and the evaluation results are intuitive and reliable, which has important scientific significance for objective analysis and evaluation of the performance of the shadow index.

[0060] The above only describes the preferred embodiments of the present application and does not limit the present application in any form. Although the present application has been disclosed as above with the preferred embodiments, it is not intended to limit the present application. Any person skilled in the art can make some changes or modifications to the above-mentioned technical content without departing from the scope of the technical solutions of the present application, and any simple modification, equivalent change and modification of the above embodiments based on the technical essence of the present application are still within the scope of the present application.

Claims

1. A method for evaluating the performance of shadow index in remote sensing images based on multiple scenarios and criteria, characterized in that the method includes the following steps: Step 1, Data Preparation and Binary Classification: Acquire multispectral image data of urban and mountain scenes, as well as digital elevation model (DEM) data corresponding to mountain scenes, and extract shadows from the images using the shadow index to be evaluated. Step 2, Sample selection: Select an appropriate number of shadow samples and non-shadow samples from the image. In urban scenes, non-shadow samples are further subdivided into six categories of easily confused land features: water bodies, vegetation, soil, light-colored impermeable surfaces, dark-colored impermeable surfaces, and blue impermeable surfaces. Non-shadow samples in mountainous scenes are not subdivided. Step 3, Separability assessment of shadows and easily confused features: Based on shadow samples and easily confused feature samples in the mountain scene, calculate the separation ratio of shadows from each type of easily confused feature at 1.5 times the quartile range (1.5IQR), and draw box plots to show the separation of shadows from each type of easily confused feature. Finally, the average 1.5IQR separation ratio of the 6 types of easily confused features is used as the 1.5IQR separation ratio of shadows from easily confused features. The 1.5IQR separation ratio is calculated using the following formula: 1.5IQRSR = 1 - 1.5IQRCR In the formula, 1.5IQRCR is the 1.5IQR confusion ratio, 1.5IQRSR is the 1.5IQR separation ratio, and Q... 3_c The upper quartile (Q3) of easily confused features, Q 1_s It is the lower quartile (Q1) of the shaded area, 1.5IQR. _c This is the 1.5 IQR range for easily confused features; Step 4, Sensitivity Assessment of Shadow Index to Illumination Intensity: Calculate the coefficient of determination R between the shadow index values ​​and the cosine of the solar incidence angle for all shadowed and non-shadowed samples in the mountain scene. 2 And plot the density scatter plot to show the change and distribution of the shadow index value with light intensity; Step 5, Shadow Extraction Accuracy Evaluation: Based on all shadow and non-shadow samples in urban and mountainous scenes, calculate the accuracy R of shadow extraction for each scene using the shadow index. a ; Step 6, Comprehensive Scoring: Summarize the quantitative calculation results from the multi-criteria evaluation and perform comprehensive scoring. The scoring results are displayed by drawing a heat map to show the performance of the shading index in each criterion.

2. The method for evaluating the performance of the shadow index of remote sensing images based on multiple scenarios and multiple criteria as described in claim 1, characterized in that, In step 5, the accuracy R of shadow extraction is determined by the shadow index. a Calculate using the following formula: In the formula, R a It is the accuracy of shadow extraction, N tds N is the number of true shadow pixels in the detected shadows. tdns N is the number of true non-shadow pixels detected in the non-shadow area. total It represents the total number of pixels in the sample.

3. The method for evaluating the shadow index performance of remote sensing images based on multiple scenarios and multiple criteria as described in claim 1, characterized in that, In step 6, the comprehensive scoring items include: the 1.5 IQR separation ratio of shadows and easily confused features in urban scenes, and the coefficient of determination R between the shadow index value and the cosine of the solar incidence angle in mountain scenes. 2 The accuracy of shadow extraction in urban and mountainous scenes is R. a The overall score is the average of all items.

Images