Multi-channel fusion image enhancement method for strengthening surface features of internal waves in remote sensing data

Abstract

The application discloses a multi-channel fusion image enhancement method for strengthening internal wave surface features in remote sensing data, adopts multiple remote sensing image channels as input information channels of the image enhancement method, fuses the internal wave surface features recorded in different channels into the same image, so that the internal wave surface features under different imaging environments can be considered, the internal wave surface features can achieve good imaging effects in the enhanced image, the information amount of the input is expanded, the internal wave surface features recorded by multiple channels are used for feature fusion by using the input information of multiple channels to further superimpose and strengthen, and finally the good strengthening effect of the internal wave surface features is realized. The method breaks through the problems of the single-channel image enhancement method for the internal wave surface features, such as the difficulty in widely adapting to the internal wave surface features under different imaging environments, and the limitation of the upper limit of the information that can be recorded and expressed by a single remote sensing imaging channel.

Description

Technical Field

[0001] This invention belongs to the field of marine engineering technology, specifically relating to a multi-channel fusion image enhancement method for enhancing internal wave surface features in remote sensing data. Background Technology

[0002] Internal waves are wave phenomena occurring at density gradients within water bodies. Internal waves have large amplitudes and carry high energy, often accompanied by strong water undulations, wave-induced current fields, and drastic changes in the distribution of seawater physical properties such as density, salinity, and temperature. These pose a significant threat to various marine structures. Based on measured data, the capture, observation, and statistical study of ocean internal waves are of great importance to the field of marine engineering.

[0003] Ocean internal waves can be observed in various ways, such as conductivity-temperature-depth profilers, acoustic Doppler current meters, echo sounders, and satellite remote sensing. Among these, satellite remote sensing imagery, with its wide field of view, has become one of the most important methods for large-scale capture and observation of internal waves. When internal waves propagate underwater, they interact with surface waves, causing convergence and divergence of the ocean surface waves above them. This results in uneven distribution of ocean surface roughness, which appears as a unique pattern of alternating bright and dark stripes in remote sensing imagery. This pattern is called the surface feature of internal waves. By observing the surface feature of internal waves, the capture and observation of ocean internal waves can be achieved.

[0004] Satellite remote sensing imagery typically has multiple channels, each with different imaging characteristics, and each is best suited for imaging different imaging environments and targets. Internal wave surface features can often only be recorded by a limited number of specific channels, and the optimal imaging channel varies depending on the imaging environment. Furthermore, because satellite remote sensing imagery often has a wide signal reception range and complex imaging content, internal wave surface features in unprocessed raw satellite remote sensing images often occupy only a very limited color range, making them difficult to observe and identify directly. Therefore, selecting appropriate channels and performing suitable image enhancement processing to highlight the imaging effect of internal wave surface features is of great significance for the discovery and observation of internal waves.

[0005] Currently, existing image enhancement methods for internal wave surface features often target a single channel. Their limitations are twofold: First, the optimal imaging channel differs for different internal wave surface features, and a single imaging channel cannot simultaneously be the optimal channel for all internal wave surface features, making it difficult to broadly adapt to internal wave surface features in different imaging environments. Selecting only a single channel as the source of information for internal wave image enhancement leads to the omission of many internal wave surface features that cannot be well imaged in the selected channel; furthermore, when the selected channel is a non-optimal imaging channel for some internal wave surface features, even if these features can be barely imaged in the selected channel, the imaging quality will be significantly inferior to that of the optimal imaging channel. Second, for any internal wave surface feature, regardless of whether a particular channel is its optimal imaging channel, the image information that a single imaging channel can record is limited, incomparable to the total information recorded by multiple channels. Using remote sensing data from only a single channel as input for image enhancement methods limits the enhancement effect to the limited amount of information input, making it impossible to break through the upper limit of information that a single remote sensing imaging channel can record and express, and thus difficult to further improve the enhancement effect. Summary of the Invention

[0006] The technical problem to be solved by the present invention is that the enhancement effect of the existing internal wave surface feature enhancement method is not ideal and the universality of internal wave surface features under different imaging environments is poor.

[0007] To address the aforementioned technical problems, the present invention provides a multi-channel fusion image enhancement method for enhancing internal wave surface features in remote sensing data, comprising the following steps: Select multiple remote sensing image channels suitable for sea surface imaging from remote sensing images, and collect remote sensing data as raw remote sensing input data. Set up an image processing channel with the same number of channels as the raw remote sensing image input data. The data recorded by each remote sensing image channel is input into one image processing channel. Color normalization is performed on the remote sensing images in each image processing channel. The image data in each channel is converted into floating-point numbers of 0-1 according to linear mapping, with each data point corresponding to a pixel. The color histogram of the normalized image in each image processing channel is calculated by dividing the floating-point number from 0 to 1 into 255 equal parts. The number of imaging pixels and the effective color range of the image are determined based on the color histogram. Based on the effective color range of the image in each image processing channel, color stretching is performed on the image of each channel respectively; Brightness equalization and color mapping are performed on the images in all image processing channels, followed by channel fusion to obtain a fused channel. The contrast of the images in the fused channel is then enhanced to obtain the final image.

[0008] Preferably, the formula corresponding to the color normalization process is as follows:

[0009] in, , These represent the row and column indices of the pixels in the remote sensing image, which are also the row and column indices of the corresponding data in the remote sensing data matrix. (Superscript) This indicates the sequence number of the currently calculated image processing channel. These are the original remote sensing image data values. These are the normalized pixel color values. , These represent the minimum and maximum values ​​that the remote sensing image data can take within the currently calculated image processing channel, respectively. (Operator) Indicates the object of its action Perform floor operations.

[0010] Preferably, determining the number of imaging pixels and the effective color range of an image based on the color histogram includes the following steps: Analyze the distribution of pixels in the color histogram; The pixel value falls on and The set of pixels within a certain range is called non-imaging pixels, and the number of pixels contained in a non-imaging pixel set is called the number of non-imaging pixels. Conversely, the set of pixels within a certain range is called imaging pixels, and the number of pixels contained in an imaging pixel set is called the number of imaging pixels. Select color stretch ratio After arranging all imaging pixels in ascending order of pixel value, the pixel with the smallest value is selected. The proportion of pixels and the largest pixel value The set of imaging pixels that are excluded from the proportion of pixels is called the effective pixels, and the number of pixels contained therein is called the number of effective pixels. The effective color range is determined by taking the minimum pixel value as the lower bound and the maximum pixel value as the upper bound.

[0011] Preferably, the color stretching uses the following formula:

[0012] in, These are pixel values ​​after color stretching. The lower limit of the effective color range, This is the upper limit of the effective color range.

[0013] Preferably, the formula used for the brightness-balanced color mapping is as follows:

[0014] in, These are the pixel values ​​of the image after brightness-equalized color mapping. Indicates channel The brightness adjustment coefficient.

[0015] Preferably, the channel Brightness adjustment coefficient The selection of the brightness level should be based on the imaging characteristics of the remote sensing image product used, taking into account the brightness characteristics of all selected channels, so that the brightness levels of each channel image after brightness equalization and color mapping are roughly similar.

[0016] Preferably, the fusion formula used to fuse the brightness-equalized color-mapped images is as follows:

[0017] in, These are the pixel values ​​of the image after brightness-equalized color mapping. These are the pixel values ​​of the merged image.

[0018] Preferably, the remote sensing image channel is selected based on the imaging characteristics of the remote sensing product used.

[0019] Preferably, the color stretching ratio during color stretching is adjusted according to the imaging characteristics of the remote sensing product used.

[0020] Preferably, the color stretching ratio is selected in the range of 1%-10%.

[0021] The present invention proposes a multi-channel fusion image enhancement method for enhancing internal wave surface features in remote sensing data. It uses multiple remote sensing image channels as input information for the image enhancement method, and fuses internal wave surface features from different channels into the same channel to achieve imaging effects that take into account internal wave surface features under different imaging environments, and expands the amount of input information. The internal wave surface feature images are superimposed and enhanced during the fusion process of multiple channels. Attached Figure Description

[0022] Figure 1 This is a flowchart illustrating the multi-channel fusion image enhancement method for enhancing internal wave surface features in remote sensing data, as provided in an embodiment of the present invention. Figure 2 A MODIS optical remote sensing true-color image of example scene A provided in this embodiment of the invention; Figure 3 The image of example scenario A provided in this embodiment of the invention is processed by image processing channels 1, 2, and 3; Figure 4 The example scene A is a remote sensing image of multi-channel fused image enhancement provided in an embodiment of the present invention. Detailed Implementation

[0023] The present invention will be further illustrated below with reference to specific embodiments. It should be understood that these embodiments are for illustrative purposes only and are not intended to limit the scope of the invention. Furthermore, it should be understood that after reading the teachings of this invention, those skilled in the art can make various alterations or modifications to the invention, and these equivalent forms also fall within the scope defined by the appended claims.

[0024] like Figure 1 As shown, this embodiment of the invention provides a multi-channel fusion image enhancement method for enhancing internal wave surface features in remote sensing data, comprising the following steps: S1: Select multiple remote sensing image channels suitable for sea surface imaging from the remote sensing image and collect remote sensing data as raw remote sensing input data. Set an image processing channel with the same number as the remote sensing image input data channels. The data recorded by each remote sensing image channel is input into one image processing channel.

[0025] S2: Perform color normalization on the remote sensing image in each image processing channel, and convert the image data in each channel into floating-point numbers of 0-1 according to linear mapping, with each data point corresponding to a pixel.

[0026] The specific method for image processing channel color normalization in step S2 is as follows:

[0027] in, , These represent the row and column indices of the pixels in the remote sensing image, which are also the row and column indices of the corresponding data in the remote sensing data matrix. (Superscript) This indicates the sequence number of the currently calculated image processing channel. These are the original remote sensing image data values. These are the pixel color values ​​after normalization. , These represent the minimum and maximum values ​​that the remote sensing image data can take within the currently calculated image processing channel, respectively. Operators Indicates the object of its action Perform floor operations.

[0028] S3: Calculate the color histogram of the image in each image processing channel, and then determine the number of imaging pixels and the effective color range of the image based on the color histogram.

[0029] The specific process for determining the number of imaging pixels and the effective color range in the image processing channel in step S3 is as follows: S3.1 Calculate the color histogram of the normalized image by dividing 0-1 into 255 equal parts; S3.2 Analyze the distribution of pixels in the color histogram and divide all pixels into two categories: those whose pixel values ​​fall within... and The set of pixels within a certain range is called the non-imaging pixel set, and the number of pixels it contains is called the non-imaging pixel count. The set of the remaining pixels is called the imaging pixel set, denoted as . The number of pixels contained in the image is called the number of imaging pixels, denoted as . ; S3.3, Select the color stretching ratio Generally, the percentage should be selected within the range of 1%-10%, with a recommended value of 5%. All imaging pixels... After sorting by pixel value from smallest to largest, select the pixel with the smallest value. The largest number of pixels and pixel values The set of remaining imaging pixels after excluding the excluded pixels is called the effective pixels. The number of pixels contained therein is called the number of effective pixels. .

[0030] S3.4, [Identify effective pixels] In, the smallest pixel value As the lower bound of the effective color range, the maximum pixel value As the upper limit of the effective color range, the range This is the effective color range.

[0031] S4: Based on the effective color range of the image in each image processing channel, perform color stretching on each channel image respectively.

[0032] The specific method for color stretching of the image in each image processing channel as described in step S4 is as follows:

[0033] in, These are the pixel values ​​after color stretching.

[0034] S5: Perform brightness equalization and color mapping on the images in all image processing channels, followed by channel fusion to obtain a fused channel. Then, enhance the contrast of the image in the fused channel to obtain the final image.

[0035] The specific process of channel fusion described in step S5 is as follows: S5.1 Perform brightness equalization color mapping on all channels. The equalization method is as follows:

[0036] in, These are the pixel values ​​of the image after brightness-equalized color mapping. Indicates channel The brightness adjustment coefficient. The range of values ​​is The smaller the value, the greater the increase in brightness, and vice versa. The selection of the brightness level should be based on the imaging characteristics of the remote sensing image product used, taking into account the brightness characteristics of all selected channels, so that the brightness levels of each channel image after brightness equalization and color mapping are roughly similar.

[0037] S5.2. The image after brightness equalization and color mapping is fused to obtain a single fused channel. The fusion method is as follows:

[0038] in, These are the pixel values ​​of the merged image.

[0039] S5.3 The contrast of the fused image can be enhanced as needed to obtain the final image.

[0040] The following is an example: This section uses an example of internal wave surface feature enhancement imagery from MODIS optical remote sensing imagery as a case study. Example scene A—a remote sensing scene of internal wave surface features captured by the MODIS remote sensing satellite in the Sea of ​​Japan on August 14, 2014—is selected as the image case for illustrating this method.

[0041] True-color remote sensing image of example scenario A, such as Figure 2 As shown in the image, the surface features of the internal waves, which appear as alternating bright and dark stripes, cannot be directly observed. This is because the remote sensing image channels used to synthesize the true-color image (i.e., the red, green, and blue light channels, corresponding to the Band 1 remote sensing channel in the 250m resolution group, the Band 2 remote sensing channel in the 500m resolution group, and the Band 1 remote sensing channel, respectively) are not the most suitable bands for sea surface observation, and the remote sensing image lacks effective image enhancement processing.

[0042] In MODIS optical remote sensing imagery, bands 1-9 of the 1km resolution group are the most sensitive for capturing ocean color and are the most suitable for observing the surface features of internal ocean waves. Adjacent bands have similar spectral ranges and imaging characteristics, while the differences in spectral range and imaging characteristics between non-adjacent bands increase with the band number. Therefore, data from bands 3, 6, and 9 of the 1km resolution group were selected at intervals as input data and fed into image processing channels 1, 2, and 3 respectively, ensuring the input information of the method is both differentiated and representative.

[0043] In remote sensing image channels 1, 2, and 3, color normalization is performed sequentially, color histograms are calculated, effective color ranges are determined, and color stretching is applied. The resulting processed images are shown below. Figure 3 As shown, (a) full-frame and partial images of channel 1; (b) full-frame and partial images of channel 2; and (c) full-frame and partial images of channel 3. The blue and red boxes mark two areas in the images where the internal wave surface features appear, located in the bright light area and the dim light area, respectively, and these areas are magnified in the partial images. Figure 3 Image processing channel 1, shown in (a), has the best dynamic range and overexposure resistance among the three channels, with a large effective imaging area and a small overexposure area. However, it has the worst sensitivity to light signals. Figure 3 (a2) shows that the image of the bright light area is obviously dark. Figure 3 The low-light area shown in (a3) ​​is directly presented as an overly dark area, and the sea surface information cannot be observed. Figure 3 Image processing channel 2, as shown in (b), has moderate latitude, overexposure resistance, and sensitivity among the three channels, with its underexposed area, overexposed area, and effective imaging area falling between those of image processing channels 1 and 3. Figure 3 In the high-light area shown in (b2), the imaging effect of image processing channel 2 is significantly better than that of image processing channel 1, and the imaging effect of the alternating bright and dark striped inner wave surface features is more ideal. However, the light signal sensitivity of image processing channel 2 is still insufficient to cope with low-light environments. Figure 3 (b3) shows that the low-light area is still too dark and the sea surface information cannot be observed. Figure 3 Image processing channel 3, as shown in (c), has the worst dynamic range and overexposure resistance among the three channels, with the largest overexposure area and the smallest effective imaging area. Figure 3 In the bright light area shown in (c2), image processing channel 3 is the only one of the three channels that is overexposed and cannot capture information about the sea surface. However, conversely, image processing channel 3 has the highest sensitivity among the three image processing channels. Figure 3In the low-light region shown in (c3), image processing channel 3 is the only channel that did not become too dark, still achieving good imaging and successfully observing the internal wave surface features in the low-light region. Therefore, under the condition that different image processing channels have different imaging characteristics, any single channel, limited by the limited input information, cannot simultaneously take into account both dynamic range and sensitivity, as well as both bright and low-light regions. Multi-channel fusion is needed to integrate the advantageous information features of different channels, so that sea surface information under different lighting environments and imaging characteristics can be well represented.

[0044] Brightness equalization and color mapping are performed on the processed images from image processing channels 1, 2, and 3. Considering the imaging and brightness characteristics of MODIS remote sensing imagery, the brightness adjustment coefficients for the three image processing channels are as follows: , , This achieves a better brightness balance. Subsequent channel blending and contrast enhancement result in the enhanced image, as shown below. Figure 4 As shown, (a) is the full-frame image; (b) is a partial image of the bright area; and (c) is a partial image of the dim area. After multi-channel fusion image enhancement, the effective imaging area of ​​the remote sensing image of example scene A is significantly expanded, and the color levels are richer. The internal wave surface features under both bright and dim light environments are taken into account and can be displayed in the image simultaneously, with both showing good performance. The method achieves image enhancement of internal wave surface features.

[0045] The alternative technical solutions are as follows: (1) In step S1, the remote sensing image channel selected as input data can be flexibly selected according to the imaging characteristics of the remote sensing product used.

[0046] (2) In step S3, the color stretching ratio during color stretching can be adjusted appropriately according to the imaging characteristics of the remote sensing product used.

[0047] (3) In step S5, the brightness adjustment coefficient when performing brightness equalization color mapping can be appropriately adjusted according to the imaging characteristics of the remote sensing product used and the brightness characteristics of different image processing channels.

[0048] The multi-channel fusion image enhancement method for enhancing internal wave surface features in remote sensing data provided in this invention overcomes the limitations of single-channel image enhancement methods for internal wave surface features, which struggle to adapt to internal wave surface features in different imaging environments and are restricted by the upper limit of information that a single remote sensing imaging channel can record and express, by simultaneously using data from multiple remote sensing image channels as input information. By fusing internal wave surface features from different channels into the same image, this method takes into account internal wave surface features under different imaging environments, ensuring good imaging results for all features. Furthermore, the internal wave surface feature images are superimposed and enhanced during the fusion process, ultimately achieving a good enhancement effect for the internal wave surface features.

Claims

1. A multi-channel fusion image enhancement method for enhancing internal wave surface features in remote sensing data, characterized in that, Includes the following steps: Select multiple remote sensing image channels suitable for sea surface imaging from remote sensing images, and collect remote sensing data as raw remote sensing input data. Set up an image processing channel with the same number of channels as the raw remote sensing image input data. The data recorded by each remote sensing image channel is input into one image processing channel. Color normalization is performed on the remote sensing images in each image processing channel. The image data in each channel is converted into floating-point numbers of 0-1 according to linear mapping, with each data point corresponding to a pixel. The color histogram of the normalized image in each image processing channel is calculated by dividing the floating-point number from 0 to 1 into 255 equal parts. The number of imaging pixels and the effective color range of the image are determined based on the color histogram. Based on the effective color range of the image in each image processing channel, color stretching is performed on the image of each channel respectively; Brightness equalization and color mapping are performed on the images in all image processing channels, followed by channel fusion to obtain a fused channel. The contrast of the images in the fused channel is then enhanced to obtain the final image.

2. The multi-channel fusion image enhancement method for enhancing internal wave surface features in remote sensing data as described in claim 1, characterized in that, The formula corresponding to the color normalization process is as follows: in, , These represent the row and column indices of the pixels in the remote sensing image, which are also the row and column indices of the corresponding data in the remote sensing data matrix. (Superscript) This indicates the sequence number of the currently calculated image processing channel. These are the original remote sensing image data values. These are the normalized pixel color values. , These represent the minimum and maximum values ​​that the remote sensing image data can take within the currently calculated image processing channel, respectively. (Operator) Indicates the object of its action Perform floor operations.

3. The multi-channel fusion image enhancement method for enhancing internal wave surface features in remote sensing data as described in claim 1, characterized in that, The process of determining the number of imaging pixels and the effective color range of an image based on a color histogram includes the following steps: Analyze the distribution of pixels in the color histogram; The pixel value falls on and The set of pixels within a certain range is called non-imaging pixels, and the number of pixels contained in a non-imaging pixel set is called the number of non-imaging pixels. Conversely, the set of pixels within a certain range is called imaging pixels, and the number of pixels contained in an imaging pixel set is called the number of imaging pixels. Select color stretch ratio After arranging all imaging pixels in ascending order of pixel value, the pixel with the smallest value is selected. The proportion of pixels and the largest pixel value The set of imaging pixels that are excluded from the proportion of pixels is called the effective pixels, and the number of pixels contained therein is called the number of effective pixels. The effective color range is determined by taking the minimum pixel value as the lower bound and the maximum pixel value as the upper bound.

4. The multi-channel fusion image enhancement method for enhancing internal wave surface features in remote sensing data as described in claim 1, characterized in that, The formula used for color stretching is as follows: in, These are pixel values ​​after color stretching. The lower limit of the effective color range, This is the upper limit of the effective color range.

5. The multi-channel fusion image enhancement method for enhancing internal wave surface features in remote sensing data as described in claim 1, characterized in that, The formula used for brightness-balanced color mapping is as follows: in, These are the pixel values ​​of the image after brightness-equalized color mapping. Indicates channel The brightness adjustment coefficient.

6. The multi-channel fusion image enhancement method for enhancing internal wave surface features in remote sensing data as described in claim 5, characterized in that, The channel Brightness adjustment coefficient The selection of the brightness level should be based on the imaging characteristics of the remote sensing image product used, taking into account the brightness characteristics of all selected channels, so that the brightness levels of each channel image after brightness equalization and color mapping are roughly similar.

7. The multi-channel fusion image enhancement method for enhancing internal wave surface features in remote sensing data as described in claim 1, characterized in that, The fusion formula used to fuse the images after brightness equalization and color mapping is as follows: in, These are the pixel values ​​of the image after brightness-equalized color mapping. These are the pixel values ​​of the merged image.

8. The multi-channel fusion image enhancement method for enhancing internal wave surface features in remote sensing data as described in claim 1, characterized in that, The remote sensing image channel is selected based on the imaging characteristics of the remote sensing product used.

9. The multi-channel fusion image enhancement method for enhancing internal wave surface features in remote sensing data as described in claim 1, characterized in that, The color stretching ratio is adjusted according to the imaging characteristics of the remote sensing product used.

10. The multi-channel fusion image enhancement method for enhancing internal wave surface features in remote sensing data as described in claim 9, characterized in that, The color stretching ratio is selected within the range of 1%-10%.

Images