A remote sensing image enhancement method and system for monitoring cultivated land planting status

Abstract

The present invention relates to the field of image enhancement technology, and in particular to a remote sensing image enhancement method and system for monitoring cultivated land planting status. The method comprises: segmenting a cultivated land remote sensing image to obtain a plurality of remote sensing sub-image blocks and clustering them, obtaining a first blur degree of the remote sensing sub-image block according to the association between the remote sensing sub-image block and other remote sensing sub-image blocks in the same cluster, thereby obtaining a suspected merged blurred area and a normal area, combining the first blur degree of the suspected merged blurred area to obtain the possibility that the suspected merged blurred area belongs to the blurred area, and then combining the distance from the center of the cultivated land remote sensing image to obtain the final possibility that the suspected merged blurred area belongs to the blurred area, thereby screening the blurred area, and performing image enhancement on the blurred area, thereby solving the blurring of the cultivated land remote sensing image caused by shooting problems and improving the accuracy of the results of cultivated land planting status detection.

Description

Technical Field

[0001] The present invention relates to the technical field of image enhancement, and in particular to a remote sensing image enhancement method and system for monitoring cultivated land planting status. Background Art

[0002] Farmland planting status monitoring uses drones to photograph farmland and help farmers understand the growth stage of crops. It helps to scientifically apply fertilizers and irrigation, effectively prevent and control pests and diseases, choose the best time to harvest, optimize crop rotation and intercropping planning, formulate market strategies, and improve the ability to cope with climate change. By mastering this information, farmers can improve production efficiency and economic benefits.

[0003] When drones are photographing cultivated land, the remote sensing images of the cultivated land appear blurred due to vibrations and jitters during flight, affecting the accuracy of subsequent detection results of the cultivated land planting status. Summary of the invention

[0004] In order to solve the technical problem that the existing remote sensing images of cultivated land will affect the accuracy of the subsequent detection results of cultivated land planting status when blurred, the purpose of the present invention is to provide a remote sensing image enhancement method and system for cultivated land planting status monitoring. The technical scheme adopted is as follows:

[0005] In a first aspect of the present invention, a remote sensing image enhancement method for monitoring cultivated land planting status is provided, comprising:

[0006] Acquire multiple remote sensing sub-image blocks obtained by segmenting the cultivated land remote sensing image, and cluster them to obtain at least two clusters;

[0007] According to the association between the edge lines and grayscale values ​​of the remote sensing sub-image block and other remote sensing sub-image blocks in the same cluster, a first blur degree of the remote sensing sub-image block is obtained;

[0008] Obtaining a suspected merged blur region and a normal region according to the first blur degree;

[0009] According to the edge shape features and grayscale values ​​of the suspected merged fuzzy region and the normal region, combined with the first blur degree corresponding to the suspected merged fuzzy region, the possibility that the suspected merged fuzzy region belongs to the fuzzy region is obtained;

[0010] According to the possibility that the suspected merged fuzzy region belongs to the fuzzy region and the distance from the center of the farmland remote sensing image, obtaining the final possibility that the suspected merged fuzzy region belongs to the fuzzy region;

[0011] According to the final possibility that the suspected merged blur region belongs to the blur region, the blur region is screened and image enhancement is performed on the blur region.

[0012] In an exemplary embodiment, obtaining a first blur degree of the remote sensing sub-image block according to the association between edge lines and grayscale values ​​of the remote sensing sub-image block and other remote sensing sub-image blocks in the same cluster includes:

[0013] Obtaining edge lines in the remote sensing sub-image block, and obtaining a fuzzy index of the remote sensing sub-image block according to the number of pixel points of the edge lines in the remote sensing sub-image block and the grayscale value variance of the pixel points of the remote sensing sub-image block;

[0014] A first blur degree of the remote sensing sub-image block is obtained according to blur indices of the remote sensing sub-image block and other remote sensing sub-image blocks in the same cluster.

[0015] In an exemplary embodiment, the calculation formula of the fuzzy index of the remote sensing sub-image block is:

[0016] ;

[0017] Among them, i represents the remote sensing sub-image blocks, Indicates The fuzzy index of a remote sensing sub-image block; Indicates In the remote sensing sub-image block edge line; Indicates The number of edge lines in a remote sensing sub-image block; Indicates In the remote sensing sub-image block The number of pixels on the edge line; Indicates The number of pixels in a remote sensing sub-image block; Indicates The variance of the grayscale values ​​of all pixels in a remote sensing sub-image block; It represents the exponential function with the natural constant e as the base;

[0018] The calculation formula of the first blur degree of the remote sensing sub-image block is:

[0019] ;

[0020] in, Indicates The first blur level of the remote sensing sub-image block; r represents the The rth remote sensing sub-image block in the cluster where the rth remote sensing sub-image block belongs; Indicates The number of remote sensing sub-image blocks in the cluster where the remote sensing sub-image block belongs; Indicates The fuzzy index of the rth remote sensing sub-image block in the cluster where the rth remote sensing sub-image block belongs.

[0021] In an exemplary embodiment, obtaining the suspected merged blur region and the normal region according to the first blur degree includes:

[0022] Taking the remote sensing sub-image block corresponding to the first blur degree greater than or equal to the preset first threshold as the suspected blur sub-region;

[0023] Adjacent suspected fuzzy sub-regions are merged into a suspected merged fuzzy region, and each remote sensing sub-image block except the suspected merged fuzzy region is used as each normal region.

[0024] In an exemplary embodiment, according to the edge shape features and grayscale values ​​of the suspected merged blurry area and the normal area, combined with the first blur degree corresponding to the suspected merged blurry area, the possibility that the suspected merged blurry area belongs to the blurry area is obtained, including:

[0025] Acquire a chain code sequence of each edge line in a first region, and obtain a first-order difference chain code absolute value sequence of the chain code sequence of each edge line; the first region is any one of the suspected merged fuzzy region and the normal region;

[0026] Obtaining the edge complexity of the first region according to the absolute value sequence of the first-order differential chain code of each edge line in the first region and the gradient amplitude of the pixel points on each edge line;

[0027] According to the difference in edge complexity between the suspected merged fuzzy region and each normal region, and the first blur degree corresponding to the suspected merged fuzzy region, the possibility that the suspected merged fuzzy region belongs to the fuzzy region is obtained.

[0028] In an exemplary embodiment, the calculation formula of the edge complexity of the first region is as follows:

[0029] ;

[0030] Among them, x represents the xth first region, Indicates the edge complexity of the xth first region; Indicates the xth first region. edge line; represents the number of edge lines in the xth first region; Indicates the xth first region. The number of data in the absolute value sequence of the first-order difference chain code of the edge line; Indicates the xth first region. The first order difference chain code absolute value sequence of the edge line individual data; Indicates the xth first region. The first order difference chain code absolute value sequence of the edge line The value of the data; Indicates the xth first region. The variance of the gradient amplitude of the pixel points on the edge line;

[0031] The calculation formula for the possibility that the suspected merged fuzzy area belongs to the fuzzy area is as follows:

[0032] ;

[0033] in, Indicates There are suspected merged fuzzy areas. Indicates The possibility that the suspected merged fuzzy regions belong to the fuzzy region; represents the normalization function; Indicates The edge complexity of the image in the suspected merged blurred area; represents the number of normal areas, Indicates Normal area; Indicates The edge complexity of a normal area; Indicates The number of remote sensing sub-image blocks contained in the suspected merged fuzzy area, z represents the The zth remote sensing sub-image block contained in the suspected merged fuzzy area; Indicates The first blur degree of the zth remote sensing sub-image block contained in the suspected merged blur area.

[0034] In an exemplary embodiment, according to the possibility that the suspected merged fuzzy region belongs to the fuzzy region and the distance from the center of the farmland remote sensing image, obtaining the final possibility that the suspected merged fuzzy region belongs to the fuzzy region includes:

[0035] Obtaining the distance between the center of each suspected merged fuzzy area and the center of the farmland remote sensing image, and sorting the possibility that each suspected merged fuzzy area belongs to the fuzzy area in ascending order of distance to obtain a possibility sequence;

[0036] Obtaining each possibility in the possibility sequence before the possibility that the suspected merged fuzzy region belongs to the fuzzy region;

[0037] According to the difference between the possibility that the suspected merged fuzzy region belongs to the fuzzy region and the previous possibilities, a final possibility that the suspected merged fuzzy region belongs to the fuzzy region is obtained.

[0038] In an exemplary embodiment, the calculation formula for the final probability that the suspected merged fuzzy region belongs to the fuzzy region is as follows:

[0039] ;

[0040] in, Indicates There are suspected merged fuzzy areas. Indicates The final probability that the suspected merged fuzzy regions belong to the fuzzy region; represents the normalization function; Indicates that in the possible sequence, The number of possibilities before the suspected merged fuzzy region belongs to the fuzzy region; Indicates that in the possible sequence, the The possibility that the suspected merged fuzzy area belongs to the fuzzy area before Possibility Indicates that in the possible sequence, the The possibility that the suspected merged fuzzy area belongs to the fuzzy area before The value of the probability; Indicates The possibility that the suspected merged fuzzy regions belong to the fuzzy region.

[0041] In an exemplary embodiment, according to the final possibility that the suspected merged blur region belongs to the blur region, the blur region is screened and image enhancement is performed on the blur region, including:

[0042] Determine the suspected merged fuzzy area corresponding to the final probability greater than or equal to the preset second threshold as the fuzzy area;

[0043] Obtaining a linear enhancement coefficient of the corresponding fuzzy area according to the final possibility of the determined fuzzy area;

[0044] The image of the blurred area is enhanced according to the linear enhancement coefficient of the blurred area.

[0045] In a second aspect of the present invention, a remote sensing image enhancement system for monitoring the planting status of cultivated land is provided, comprising: a memory and a processor; the memory is connected to the processor; the memory is used to store program instructions; the processor is used to implement the above-mentioned remote sensing image enhancement method for monitoring the planting status of cultivated land when the program instructions are executed.

[0046] The present invention has the following beneficial effects: firstly, the remote sensing image of cultivated land is segmented to obtain a plurality of remote sensing sub-image blocks, and clustered to obtain at least two clusters for convenience of subsequent processing; then, the fuzzy area in the remote sensing image of cultivated land is screened out according to the relevant features of the fuzzy area, as well as the distinction and association with other areas, so as to enhance the image of the fuzzy area according to the degree of fuzziness of the fuzzy area, solve the fuzziness of the remote sensing image of cultivated land caused by shooting problems, and improve the accuracy of the results of the detection of the planting status of cultivated land. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 A flowchart of a remote sensing image enhancement method for monitoring cultivated land planting status provided by an embodiment of the present invention;

[0048] Figure 2 is a flowchart for obtaining the first fuzziness level;

[0049] Figure 3 It is a flow chart for obtaining the possibility that the suspected merged fuzzy region belongs to the fuzzy region;

[0050] Figure 4 It is a flowchart for obtaining the final possibility that the suspected merged fuzzy area belongs to the fuzzy area. DETAILED DESCRIPTION

[0051] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation methods, structures, features and effects of the present invention are described in detail below in conjunction with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0052] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0053] The present embodiment provides a remote sensing image enhancement method for monitoring the planting status of cultivated land. The application scenario of the remote sensing image enhancement method is: using a drone to photograph cultivated land to obtain remote sensing images of cultivated land. Due to some special reasons during the flight of the drone, the captured remote sensing images of cultivated land may appear blurred, such as: the drone vibrates due to its own reasons or external environmental reasons.

[0054] The overall concept of a remote sensing image enhancement method for monitoring the planting status of cultivated land provided in this embodiment is as follows: since the blur degree of different areas in the cultivated land remote sensing image may be different, the cultivated land remote sensing image is first divided into several areas, and clustered according to the gray value mean of all pixels in the area, and similar areas are divided into a cluster; the clearer the area, the more details are retained, and the greater the proportion of edge lines in the area, and compared with other areas of the cluster, so as to calculate the first blur degree of each area. Then, several suspected merged fuzzy areas and normal areas are obtained. When edge detection is performed on the fuzzy area, there will be some pseudo-edge lines on the edge line, which are usually irregular, while the edges of crops in the cultivated land will be relatively smooth, so as to calculate the edge complexity of all edge lines of each suspected merged fuzzy area and normal area, and compare the edge complexity of each suspected merged fuzzy area with all normal areas. The greater the difference, the greater the possibility that the suspected merged fuzzy area is a fuzzy area. In this way, the possibility that each suspected merged fuzzy area is a fuzzy area is calculated. In order to improve the accuracy of blurred area screening, further screening is required in combination with the blur degree and position relationship of the blurred area: since the lens used by drones usually has higher optical quality in the central area, the center of the cultivated land remote sensing image will be clearer, while the edge is relatively blurred. Based on this feature, the final possibility that each suspected merged blurred area is a blurred area is calculated, and then several blurred areas are screened out, and the blurred areas are enhanced to different degrees to obtain the enhanced image.

[0055] The purpose of this embodiment to provide a remote sensing image enhancement method for monitoring the planting status of cultivated land is to screen out the fuzzy areas in the remote sensing image of cultivated land according to the characteristics of the fuzzy areas, and then enhance the fuzzy areas accordingly according to the degree of blurriness of the fuzzy areas, thereby improving the accuracy of the results of the detection of the planting status of cultivated land and avoiding misjudgment of the recognition results due to image blur.

[0056] like Figure 1 As shown, the remote sensing image enhancement method for monitoring the cultivated land planting status includes the following steps:

[0057] Step 1: Obtain multiple remote sensing sub-image blocks obtained by segmenting the cultivated land remote sensing image, and cluster them to obtain at least two clusters.

[0058] When obtaining remote sensing images of cultivated land through drones, it should be understood that the height of the drone is determined by the actual situation, such as the growth stage of the crops on the cultivated land. If the remote sensing image of cultivated land is not a grayscale image, it needs to be converted into a grayscale image. In an exemplary embodiment, the remote sensing image of cultivated land can also be processed using mean filtering to eliminate noise in the image. Among them, mean filtering is a conventional technology, and the specific method is not introduced here.

[0059] In an exemplary embodiment, the cultivated land remote sensing image is divided equally to obtain a plurality of remote sensing sub-image blocks of equal size. The size of the remote sensing sub-image block is based on the ability to divide the cultivated land remote sensing image equally. In an exemplary embodiment, the size of the remote sensing sub-image block is 10*10. In this embodiment, a two-dimensional coordinate system is constructed according to the length and width of the cultivated land remote sensing image, and the cultivated land remote sensing image is mapped to the two-dimensional coordinate system. The position of each pixel point of the cultivated land remote sensing image in the two-dimensional coordinate system is used as the two-dimensional coordinate point of each pixel point.

[0060] After obtaining multiple remote sensing sub-image blocks, cluster the multiple remote sensing sub-image blocks to obtain at least two clusters. The clustering algorithm is set according to actual needs. In an exemplary embodiment, the gray value mean of the pixel points of the remote sensing sub-image block is calculated, and the k-means algorithm is used to cluster all remote sensing sub-image blocks according to the gray value mean of each remote sensing sub-image block, wherein the preset clustering number k can be set to 2, and finally 2 clusters are obtained. Among them, the k-means algorithm is a conventional technology, and the specific method is not introduced here.

[0061] Step 2: According to the association between the edge lines and grayscale values ​​of the remote sensing sub-image block and other remote sensing sub-image blocks in the same cluster, a first blur degree of the remote sensing sub-image block is obtained.

[0062] It should be noted that: if the remote sensing sub-image block is clearer, the more details it retains, the greater the proportion of edge lines in the remote sensing sub-image block, and when the variance of the grayscale values ​​of the pixels in the remote sensing sub-image block is higher, it means that the difference between the grayscale values ​​of the pixels in the remote sensing sub-image block is larger, and the details and edges in the remote sensing sub-image block are more obvious. Therefore, according to the above characteristics, the fuzzy index of each remote sensing sub-image block is calculated. And the fuzzy index of each remote sensing sub-image block is compared with the fuzzy index of other remote sensing sub-image blocks in the same cluster, and the first fuzzy degree of each remote sensing sub-image block is calculated.

[0063] In an exemplary embodiment, Figure 2 As shown, the process of obtaining the first blur level includes:

[0064] Step 2-1: Obtain edge lines in the remote sensing sub-image block, and obtain the fuzzy index of the remote sensing sub-image block according to the number of pixel points of the edge lines in the remote sensing sub-image block and the gray value variance of the pixel points of the remote sensing sub-image block.

[0065] Taking any remote sensing sub-image block as an example, an edge detection algorithm is used to obtain edge lines in the remote sensing sub-image block. The edge detection algorithm is a conventional technology and the specific method is not introduced here.

[0066] According to the number of edge line pixels in the remote sensing sub-image block and the grayscale value variance of the pixels in the remote sensing sub-image block, the fuzzy index of the remote sensing sub-image block is obtained. The more the number of edge line pixels in the remote sensing sub-image block, the smaller the fuzzy index of the remote sensing sub-image block; the larger the grayscale value variance of the pixels in the remote sensing sub-image block, the clearer the remote sensing sub-image block, and the smaller the fuzzy index of the remote sensing sub-image block. Therefore, in an exemplary embodiment, the calculation formula of the fuzzy index of the remote sensing sub-image block is:

[0067] ;

[0068] Among them, i represents the remote sensing sub-image blocks, Indicates The fuzzy index of a remote sensing sub-image block; Indicates In the remote sensing sub-image block edge line; Indicates The number of edge lines in a remote sensing sub-image block; Indicates In the remote sensing sub-image block The number of pixels on the edge line; Indicates The number of pixels in a remote sensing sub-image block; Indicates The variance of the grayscale values ​​of all pixels in a remote sensing sub-image block; Represents an exponential function with the natural constant e as the base.

[0069] Indicates The proportion of edge line pixels in the remote sensing sub-image block is larger. The smaller the fuzzy index of the remote sensing sub-image block is.

[0070] By adopting the above process, the fuzzy index of each remote sensing sub-image block is obtained.

[0071] Step 2-2: Obtain a first blur degree of the remote sensing sub-image block according to blur indices of the remote sensing sub-image block and other remote sensing sub-image blocks in the same cluster.

[0072] Taking any remote sensing sub-image block as an example, obtaining the cluster containing the remote sensing sub-image block, then each remote sensing sub-image block in the cluster can be obtained. The remote sensing sub-image blocks in the cluster where the remote sensing sub-image blocks belong include remote sensing sub-image blocks, that is, for the For a cluster of remote sensing sub-image blocks, the cluster contains remote sensing sub-image blocks, as well as other remote sensing sub-image blocks in the same cluster.

[0073] According to the blur index of the remote sensing sub-image block and other remote sensing sub-image blocks in the same cluster, the first blur degree of the remote sensing sub-image block is obtained. In an exemplary embodiment, the calculation formula of the first blur degree of the remote sensing sub-image block is:

[0074] ;

[0075] in, Indicates The first blur level of the remote sensing sub-image block; r represents the The rth remote sensing sub-image block in the cluster where the rth remote sensing sub-image block belongs; Indicates The number of remote sensing sub-image blocks in the cluster where the remote sensing sub-image block belongs; Indicates The fuzzy index of the rth remote sensing sub-image block in the cluster where the rth remote sensing sub-image block belongs.

[0076] Since the fuzzy index of each remote sensing sub-image block is obtained in the same way, The fuzzy index of the rth remote sensing sub-image block in the cluster where the remote sensing sub-image block belongs The calculation process is similar to Fuzzy index of remote sensing sub-image blocks The calculation process is the same.

[0077] The denominator in the above formula is The average value of the fuzzy index of the remote sensing sub-image blocks in the cluster where the remote sensing sub-image blocks are located indicates the overall level of the fuzzy index corresponding to the cluster.

[0078] Indicates The difference between the fuzzy index of the remote sensing sub-image block and the overall level of the fuzzy index in the cluster to which it belongs. The greater the fuzzy index of a remote sensing sub-image block is than the average value of the fuzzy index in the cluster, the The higher the fuzzy index of a remote sensing sub-image block is than the overall level of the fuzzy index in the cluster, the The more blurred the remote sensing sub-image block is, the The smaller the fuzzy index of a remote sensing sub-image block is than the average value of the fuzzy index in the cluster, the The lower the fuzzy index of the remote sensing sub-image block is than the overall level of the fuzzy index in the cluster. Then, the ratio may be greater than 1, equal to 1, or less than 1. The larger the value, the lower the fuzzy index of the remote sensing sub-image block is. The greater the first blur degree of the remote sensing sub-image block.

[0079] By adopting the above process, the first blur degree of each remote sensing sub-image block is obtained.

[0080] Step 3: Obtain the suspected merged blurred area and the normal area according to the first blur degree.

[0081] A first threshold is preset, and the first threshold is used to compare with the first blur level of each remote sensing sub-image block. The specific value of the first threshold is set according to actual needs, for example, the first threshold is 0.8.

[0082] The first blur degree of each remote sensing sub-image block is compared with the first threshold, and the remote sensing sub-image block corresponding to the first blur degree greater than or equal to the first threshold is taken as a suspected blur sub-region, thereby obtaining multiple suspected blur sub-regions.

[0083] Among the various suspected fuzzy sub-regions obtained, some of them may be adjacent, that is, there are adjacent pixels in the boundaries of two suspected fuzzy sub-regions. Then, the adjacent suspected fuzzy sub-regions are merged into a suspected merged fuzzy region. For example, if there are three suspected fuzzy sub-regions adjacent, these three suspected fuzzy sub-regions are merged into one suspected merged fuzzy region. It should be noted that there may be an isolated suspected fuzzy sub-region, that is, there are no other suspected fuzzy sub-regions around it, then the isolated suspected fuzzy sub-region is regarded as a suspected merged fuzzy region. Thus, multiple suspected merged fuzzy regions are obtained.

[0084] It should be understood that, in each remote sensing sub-image block, in addition to the remote sensing sub-image block divided into the suspected merged fuzzy area, there may be other remote sensing sub-image blocks, and these remote sensing sub-image blocks other than the suspected merged fuzzy area are regarded as normal areas. Therefore, the normal area is not merged, and in each remote sensing sub-image block other than the suspected merged fuzzy area, one remote sensing sub-image block is a normal area.

[0085] Thus, various suspected merged fuzzy areas and normal areas in the cultivated land remote sensing image are obtained.

[0086] Step 4: Based on the edge shape features and grayscale values ​​of the suspected merged fuzzy region and the normal region, combined with the first blur degree corresponding to the suspected merged fuzzy region, the possibility that the suspected merged fuzzy region belongs to the fuzzy region is obtained.

[0087] Through the above steps, each suspected merged fuzzy area and each normal area in the cultivated land remote sensing image are obtained. When edge detection is performed on the fuzzy area, there will be some pseudo edge lines on the edge line, which are usually irregular, while the edges of crops in the cultivated land are relatively smooth. In this way, the edge complexity of all edge lines of each suspected merged fuzzy area and normal area is calculated, and the edge complexity of each suspected merged fuzzy area is compared with all normal areas. The greater the difference, the greater the possibility that the suspected merged fuzzy area is a fuzzy area. In this way, the possibility of each suspected merged fuzzy area being a fuzzy area is calculated.

[0088] In an exemplary embodiment, the process of obtaining the possibility that the suspected merged fuzzy region belongs to the fuzzy region is as follows: Figure 3 As shown, including:

[0089] Step 4-1: Obtain the chain code sequence of each edge line in the first region, and obtain the first-order difference chain code absolute value sequence of the chain code sequence of each edge line.

[0090] Since the edge complexity of each suspected merged blur region and each normal region needs to be obtained later, for the convenience of description, the first region is set to be any one of the suspected merged blur region and the normal region.

[0091] Get each edge line in the first area, traverse from any endpoint of each edge line as the starting point, get the chain code sequence of each edge line, and then get the first-order differential chain code absolute value sequence of the chain code sequence of each edge line. It should be noted that the 8-chain code is used here to obtain the chain code sequence, and the chain code sequence acquisition method is a conventional technology, and the specific method is not introduced here. Each data in the first-order differential chain code absolute value sequence is the absolute value of the difference between two adjacent data in the chain code sequence, thereby forming a first-order differential chain code absolute value sequence.

[0092] Step 4-2: Obtain the edge complexity of the first region according to the absolute value sequence of the first-order differential chain code of each edge line in the first region and the gradient amplitude of the pixel points on each edge line.

[0093] For any edge line in the first region, the gradient amplitude of each pixel on the edge line is obtained, and then the variance of the gradient amplitude of the pixel on the edge line is calculated. The larger the variance of the gradient amplitude of the pixel on the edge line, the more complex and less smooth the shape of the edge line is.

[0094] According to the absolute value sequence of the first-order differential chain code of each edge line in the first region and the gradient amplitude of the pixel points on each edge line, the edge complexity of the first region is obtained. In an exemplary embodiment, the calculation formula of the edge complexity of the first region is as follows:

[0095] ;

[0096] Among them, x represents the xth first region, Indicates the edge complexity of the xth first region; Indicates the xth first region. edge line; represents the number of edge lines in the xth first region; Indicates the xth first region. The number of data in the absolute value sequence of the first-order difference chain code of the edge line; Indicates the xth first region. The first order difference chain code absolute value sequence of the edge line individual data; Indicates the xth first region. The first order difference chain code absolute value sequence of the edge line The value of the data; Indicates the xth first region. The variance of the gradient amplitude of the pixel points on the edge line.

[0097] Therefore, the xth first region The larger the variance of the gradient amplitude of the pixel points on the edge line, the greater the The more complex the shape of the edge line is, the greater the edge complexity of the x-th first region is; The larger the value, the The suspected merged fuzzy area The larger the absolute value sequence of the first-order difference chain code corresponding to the edge line is, the greater the fluctuation of the corresponding chain code sequence is, which means that the The suspected merged fuzzy area The more complex the shape of the edge line is, the greater the edge complexity of the x-th first region is.

[0098] Step 4-3: According to the difference in edge complexity between the suspected merged fuzzy region and each normal region, and the first blur degree corresponding to the suspected merged fuzzy region, the possibility that the suspected merged fuzzy region belongs to the fuzzy region is obtained.

[0099] Normally, the edge complexity of the suspected merged blurred area is greater than that of the normal area because the edge lines are more complex, irregular and blurred, and there is a large difference in the edge complexity between the suspected merged blurred area and the normal area. Therefore, the greater the difference in edge complexity between the suspected merged blurred area and the normal area, the greater the possibility that the suspected merged blurred area is a blurred area.

[0100] Since the suspected merged fuzzy area is composed of at least one remote sensing sub-image block, and each remote sensing sub-image block has a first degree of blur, therefore, each remote sensing sub-image block merged into the suspected merged fuzzy area is obtained, that is, each remote sensing sub-image block contained in the suspected merged fuzzy area, and the first degree of blur of these remote sensing sub-image blocks is obtained as the first degree of blur corresponding to the suspected merged fuzzy area. Then, the greater the first degree of blur corresponding to the suspected merged fuzzy area, the greater the possibility that the suspected merged fuzzy area belongs to the fuzzy area. Therefore, according to the difference in edge complexity between the suspected merged fuzzy area and each normal area, and the first degree of blur corresponding to the suspected merged fuzzy area, the possibility that the suspected merged fuzzy area belongs to the fuzzy area is obtained. In an exemplary embodiment, the calculation formula for the possibility that the suspected merged fuzzy area belongs to the fuzzy area is as follows:

[0101] ;

[0102] in, Indicates There are suspected merged fuzzy areas. Indicates The possibility that the suspected merged fuzzy regions belong to the fuzzy region; Indicates The edge complexity of the image in the suspected merged blurred area; represents the number of normal areas, Indicates Normal area; Indicates The edge complexity of a normal area; Indicates The number of remote sensing sub-image blocks contained in the suspected merged fuzzy area, z represents the The zth remote sensing sub-image block contained in the suspected merged fuzzy area; Indicates The first blur degree of the zth remote sensing sub-image block contained in the suspected merged blur area. Represents the normalization function, which can be a linear normalization method or the following normalization method: , where n is the input data.

[0103] Indicates The difference between the average value of edge complexity of the suspected merged fuzzy area and the normal area is greater. The greater the possibility that the suspected merged fuzzy area belongs to the fuzzy area. Indicates The average value of the first blur degree of the remote sensing sub-image blocks contained in the suspected merged blur region, that is, the overall level. The larger the average value, the The greater the possibility that the suspected merged fuzzy area belongs to the fuzzy area.

[0104] By using the above process, the possibility that each suspected merged fuzzy area belongs to the fuzzy area is obtained.

[0105] Step 5: According to the possibility that the suspected merged fuzzy area belongs to the fuzzy area and the distance from the center of the cultivated land remote sensing image, the final possibility that the suspected merged fuzzy area belongs to the fuzzy area is obtained.

[0106] Through step 4, the possibility of each suspected merged fuzzy area belonging to the fuzzy area is obtained. In order to improve the accuracy of fuzzy area screening, further screening is performed in combination with the distance between the suspected merged fuzzy area and the center of the farmland remote sensing image. The reason why the distance between the suspected merged fuzzy area and the center of the farmland remote sensing image is also combined is that the lens used by the drone usually has a higher optical quality in the center area of ​​the image, resulting in the center position of the farmland remote sensing image being clearer, and the farther away from the center, the more blurred the image. Based on the above characteristics, the final possibility of the suspected merged fuzzy area belonging to the fuzzy area is obtained.

[0107] In an exemplary embodiment, the process of obtaining the final probability that the suspected merged fuzzy region belongs to the fuzzy region is as follows: Figure 4 As shown, including:

[0108] Step 5-1: Obtain the distance between the center of each suspected merged fuzzy area and the center of the farmland remote sensing image, and sort the possibility of each suspected merged fuzzy area belonging to the fuzzy area in order of distance from small to large, and obtain a possibility sequence.

[0109] The center of each suspected merged fuzzy area is obtained. In this embodiment, the mean of the horizontal coordinates of the two-dimensional coordinate points of all the pixels of the suspected merged fuzzy area and the mean of the vertical coordinates of the two-dimensional coordinate points of all the pixels of the suspected merged fuzzy area are calculated, and the horizontal coordinates and vertical coordinates of the center of the suspected merged fuzzy area are respectively used as the horizontal coordinates and vertical coordinates of the center of the suspected merged fuzzy area to obtain the center position of the suspected merged fuzzy area. The center of the cultivated land remote sensing image is also obtained.

[0110] Then, the distance between the center of each suspected merged fuzzy area and the center of the cultivated land remote sensing image is obtained, that is, the distance between the two-dimensional coordinate point of the center of each suspected merged fuzzy area and the two-dimensional coordinate point of the center of the cultivated land remote sensing image. Then, each suspected merged fuzzy area obtains a distance. The distances of each suspected merged fuzzy area are sorted from small to large. It should be understood that if there are multiple suspected merged fuzzy areas with equal distances, the order between these multiple suspected merged fuzzy areas with equal distances can be set at will.

[0111] Then, the possibility of each suspected merged fuzzy area belonging to the fuzzy area is sorted in order of distance from small to large, and a possibility sequence is obtained. Then, the suspected merged fuzzy area corresponding to the first possibility in the possibility sequence has the smallest distance from the center of the cultivated land remote sensing image, and the suspected merged fuzzy area corresponding to the last possibility in the possibility sequence has the largest distance from the center of the cultivated land remote sensing image.

[0112] Step 5-2: Obtain each possibility in the possibility sequence before the possibility that the suspected merged fuzzy region belongs to the fuzzy region.

[0113] For any suspected merged fuzzy region, according to the position of the possibility that the suspected merged fuzzy region belongs to the fuzzy region in the possibility sequence, each possibility before the possibility that the suspected merged fuzzy region belongs to the fuzzy region in the possibility sequence can be obtained, and the suspected merged fuzzy region corresponding to each possibility before the possibility that the suspected merged fuzzy region belongs to the fuzzy region is closer to the center of the cultivated land remote sensing image. For example: if the possibility that the suspected merged fuzzy region belongs to the fuzzy region ranks fourth in the possibility sequence, then the first three possibilities in the possibility sequence are obtained.

[0114] Step 5-3: According to the difference between the possibility that the suspected merged fuzzy region belongs to the fuzzy region and its previous possibilities, the final possibility that the suspected merged fuzzy region belongs to the fuzzy region is obtained.

[0115] According to the difference between the possibility that the suspected merged fuzzy area belongs to the fuzzy area and the previous possibilities, the final possibility that the suspected merged fuzzy area belongs to the fuzzy area is obtained, wherein the greater the difference, the more consistent it is: the farther away from the center of the cultivated land remote sensing image, the more blurred the image, and the greater the final possibility that the suspected merged fuzzy area belongs to the fuzzy area. In an exemplary embodiment, the calculation formula of the final possibility that the suspected merged fuzzy area belongs to the fuzzy area is as follows:

[0116] ;

[0117] in, Indicates The final probability that the suspected merged fuzzy regions belong to the fuzzy region; In the possible sequence, The number of possibilities before the suspected merged fuzzy region belongs to the fuzzy region; In the possibility sequence, The possibility that the suspected merged fuzzy area belongs to the fuzzy area before Possibility In the possibility sequence, The possibility that the suspected merged fuzzy area belongs to the fuzzy area before The value of the probability; Indicates The possibility that the suspected merged fuzzy regions belong to the fuzzy region.

[0118] It should be understood that for the first possibility in the possibility sequence, since there is no previous possibility, the first possibility can be directly used as its corresponding final possibility.

[0119] Using the above process, the final probability that each suspected merged fuzzy region belongs to the fuzzy region is obtained.

[0120] Step 6: According to the final possibility that the suspected merged blur area belongs to the blur area, the blur area is screened and image enhancement is performed on the blur area.

[0121] In an exemplary embodiment, a second threshold is preset, and the second threshold is used to compare with the final possibility that each suspected merged blur region belongs to a blur region. The specific value of the second threshold is set according to actual needs, such as 0.7.

[0122] Compare the second threshold with the final probability that each suspected merged blur region belongs to a blur region, obtain the final probability greater than or equal to the second threshold, thereby obtaining the suspected merged blur region corresponding to the final probability greater than or equal to the second threshold, and determine the suspected merged blur region corresponding to the final probability greater than or equal to the second threshold as a blur region. Subsequently, focus on image enhancement of the blur region.

[0123] In an exemplary embodiment, a linear enhancement algorithm is used for image enhancement. The linear enhancement algorithm is a conventional technology, and the specific method is not introduced here. As other implementations, other existing image enhancement algorithms can also be used to perform image enhancement processing on the blurred area according to actual needs.

[0124] The linear enhancement coefficient of the corresponding fuzzy area is obtained according to the final possibility of the determined fuzzy area. The final possibilities of different fuzzy areas may be different, so the linear enhancement coefficients of different fuzzy areas may also be different. In an exemplary embodiment, for any fuzzy area, the final possibility of the fuzzy area is added to the value 1, and the sum is used as the linear enhancement coefficient of the fuzzy area.

[0125] Finally, according to the linear enhancement coefficient of each blurred area, a linear enhancement algorithm is used to enhance the image of each blurred area. It should be noted that the linear enhancement coefficient is used as the slope of the linear enhancement, and the preset intercept threshold is the intercept of the linear enhancement. Among them, the preset intercept threshold can be set to 0, which is described as an example.

[0126] It should be understood that, for areas other than the blurred areas in the cultivated land remote sensing image, since these areas are not the processing objects of the technical solution of the present invention, they are not specifically limited, for example, image enhancement may not be performed.

[0127] This embodiment also provides a remote sensing image enhancement system for monitoring the planting status of cultivated land, including: a memory and a processor; the memory is connected to the processor, and the memory is used to store program instructions; the processor is used to implement the steps in the above-mentioned remote sensing image enhancement method embodiment for monitoring the planting status of cultivated land when the program instructions are executed.

[0128] In an exemplary embodiment, the present invention provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps in the above-mentioned remote sensing image enhancement method embodiment for monitoring the planting status of cultivated land.

[0129] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0130] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. A remote sensing image enhancement method for monitoring cultivated land planting status, characterized in that: include: Acquire multiple remote sensing sub-image blocks obtained by segmenting the cultivated land remote sensing image, and cluster them to obtain at least two clusters; According to the association between the edge lines and grayscale values ​​of the remote sensing sub-image block and other remote sensing sub-image blocks in the same cluster, a first blur degree of the remote sensing sub-image block is obtained; Obtaining a suspected merged blur region and a normal region according to the first blur degree; According to the edge shape features and grayscale values ​​of the suspected merged fuzzy region and the normal region, combined with the first blur degree corresponding to the suspected merged fuzzy region, the possibility that the suspected merged fuzzy region belongs to the fuzzy region is obtained; According to the possibility that the suspected merged fuzzy region belongs to the fuzzy region and the distance from the center of the farmland remote sensing image, obtaining the final possibility that the suspected merged fuzzy region belongs to the fuzzy region; According to the final possibility that the suspected merged blur region belongs to the blur region, the blur region is screened and image enhancement is performed on the blur region; The acquisition of the first blur level of the remote sensing sub-image block includes: Obtaining edge lines in the remote sensing sub-image block, and obtaining a fuzzy index of the remote sensing sub-image block according to the number of pixel points of the edge lines in the remote sensing sub-image block and the grayscale value variance of the pixel points of the remote sensing sub-image block; Obtaining a first blur degree of the remote sensing sub-image block according to blur indices of the remote sensing sub-image block and other remote sensing sub-image blocks in the same cluster; The possibility that the suspected merged fuzzy area belongs to the fuzzy area is obtained, including: Acquire a chain code sequence of each edge line in a first region, and obtain a first-order difference chain code absolute value sequence of the chain code sequence of each edge line; the first region is any one of the suspected merged fuzzy region and the normal region; Obtaining the edge complexity of the first region according to the absolute value sequence of the first-order differential chain code of each edge line in the first region and the gradient amplitude of the pixel points on each edge line; According to the difference in edge complexity between the suspected merged fuzzy region and each normal region, and the first blur degree corresponding to the suspected merged fuzzy region, the possibility that the suspected merged fuzzy region belongs to the fuzzy region is obtained.

2. A remote sensing image enhancement method for monitoring cultivated land planting status as claimed in claim 1, characterized in that: The calculation formula of the fuzzy index of the remote sensing sub-image block is: ； Among them, i represents the remote sensing sub-image blocks, Indicates The fuzzy index of a remote sensing sub-image block; Indicates In the remote sensing sub-image block edge line; Indicates The number of edge lines in a remote sensing sub-image block; Indicates In the remote sensing sub-image block The number of pixels on the edge line; Indicates The number of pixels in a remote sensing sub-image block; Indicates The variance of the grayscale values ​​of all pixels in a remote sensing sub-image block; It represents the exponential function with the natural constant e as the base; The calculation formula of the first blur degree of the remote sensing sub-image block is: ； in, Indicates The first blur level of the remote sensing sub-image block; r represents the The rth remote sensing sub-image block in the cluster where the rth remote sensing sub-image block belongs; Indicates The number of remote sensing sub-image blocks in the cluster where the remote sensing sub-image block belongs; Indicates The fuzzy index of the rth remote sensing sub-image block in the cluster where the rth remote sensing sub-image block belongs.

3. A remote sensing image enhancement method for monitoring cultivated land planting status as claimed in claim 1, characterized in that: The suspected merged blur area and the normal area are obtained according to the first blur level, including: Taking the remote sensing sub-image block corresponding to the first blur degree greater than or equal to the preset first threshold as the suspected blur sub-region; Adjacent suspected fuzzy sub-regions are merged into a suspected merged fuzzy region, and each remote sensing sub-image block except the suspected merged fuzzy region is used as each normal region.

4. A remote sensing image enhancement method for monitoring cultivated land planting status as claimed in claim 1, characterized in that: The calculation formula for the edge complexity of the first region is as follows: ； Among them, x represents the xth first region, Indicates the edge complexity of the xth first region; Indicates the xth first region. edge line; represents the number of edge lines in the xth first region; Indicates the xth first region. The number of data in the absolute value sequence of the first-order difference chain code of the edge line; Indicates the xth first region. The first order difference chain code absolute value sequence of the edge line individual data; Indicates the xth first region. The first order difference chain code absolute value sequence of the edge line The value of the data; Indicates the xth first region. The variance of the gradient amplitude of the pixel points on the edge line; The calculation formula for the possibility that the suspected merged fuzzy area belongs to the fuzzy area is as follows: ； in, Indicates There are suspected merged fuzzy areas. Indicates The possibility that the suspected merged fuzzy regions belong to the fuzzy region; represents the normalization function; Indicates The edge complexity of the image in the suspected merged blurred area; represents the number of normal areas, Indicates Normal area; Indicates The edge complexity of a normal area; Indicates The number of remote sensing sub-image blocks contained in the suspected merged fuzzy area, z represents the The zth remote sensing sub-image block contained in the suspected merged fuzzy area; Indicates The first blur degree of the zth remote sensing sub-image block contained in the suspected merged blur area.

5. The remote sensing image enhancement method for monitoring cultivated land planting status according to claim 1, characterized in that: According to the possibility that the suspected merged fuzzy area belongs to the fuzzy area and the distance from the center of the farmland remote sensing image, the final possibility that the suspected merged fuzzy area belongs to the fuzzy area is obtained, including: Obtaining the distance between the center of each suspected merged fuzzy area and the center of the farmland remote sensing image, and sorting the possibility that each suspected merged fuzzy area belongs to the fuzzy area in ascending order of distance to obtain a possibility sequence; Obtaining each possibility in the possibility sequence before the possibility that the suspected merged fuzzy region belongs to the fuzzy region; According to the difference between the possibility that the suspected merged fuzzy region belongs to the fuzzy region and the previous possibilities, a final possibility that the suspected merged fuzzy region belongs to the fuzzy region is obtained.

6. A remote sensing image enhancement method for monitoring cultivated land planting status as claimed in claim 5, characterized in that: The calculation formula for the final probability that the suspected merged fuzzy area belongs to the fuzzy area is as follows: ； in, Indicates There are suspected merged fuzzy areas. Indicates The final probability that the suspected merged fuzzy regions belong to the fuzzy region; represents the normalization function; Indicates that in the possible sequence, The number of possibilities before the suspected merged fuzzy region belongs to the fuzzy region; Indicates that in the possible sequence, the The possibility that the suspected merged fuzzy area belongs to the fuzzy area before Possibilities Indicates that in the possible sequence, the The possibility that the suspected merged fuzzy area belongs to the fuzzy area before The value of the probability; Indicates The possibility that the suspected merged fuzzy regions belong to the fuzzy region.

7. The remote sensing image enhancement method for monitoring the planting status of cultivated land according to claim 1, characterized in that: According to the final possibility that the suspected merged blur region belongs to the blur region, the blur region is screened and image enhancement is performed on the blur region, including: Determine the suspected merged fuzzy area corresponding to the final probability greater than or equal to the preset second threshold as the fuzzy area; Obtaining a linear enhancement coefficient of the corresponding fuzzy area according to the final possibility of the determined fuzzy area; The image of the blurred area is enhanced according to the linear enhancement coefficient of the blurred area.

8. A remote sensing image enhancement system for monitoring cultivated land planting status, characterized in that it includes: Memory and processor; The memory is connected to the processor; The memory is used to store program instructions; The processor is used to implement the remote sensing image enhancement method for monitoring the planting status of cultivated land described in any one of claims 1 to 7 when the program instructions are executed.

Images