CN117690024A - An integrated remote sensing identification method for rice fields with multiple planting patterns - Google Patents

Abstract

The invention relates to an integrated remote sensing identification method for rice fields with multiple planting patterns, which includes: acquiring elevation and slope image data of the study area and backscattering intensity data of ground objects in time series throughout the year; collecting samples of rice and non-rice categories points as training samples; obtain rice phenology data; calculate and extract features based on the backscattering intensity data of ground objects in the time series throughout the year; obtain combined multi-band characteristic image data; based on multi-band characteristic image data, rice and non-rice categories At the sample points, a random forest classifier is used to identify rice. The beneficial effects of the present invention are: the present invention uses SAR images to avoid the impact of weather conditions on image imaging quality and subsequent rice identification and monitoring; and, the present invention realizes the distinction between rice and non-rice through statistical characteristics, and on this basis, through Phenological characteristics realize the differentiation of rice under various planting modes and obtain more accurate and precise remote sensing monitoring results of rice fields.

Description

An integrated remote sensing identification method for rice fields with multiple planting patterns

Technical field

The present invention relates to the technical field of remote sensing images, and more specifically, to an integrated remote sensing identification method for rice fields with multiple planting patterns.

Background technique

Timely and accurate knowledge of the spatial and temporal distribution of rice plays a crucial role in supporting the government in formulating land use and food policies and ensuring sustainable food supply.

Optical and synthetic aperture radar (SAR) images are commonly used for rice identification. However, effective optical remote sensing data are often limited in cloudy and rainy areas and can only reflect rice canopy information, limiting the accuracy of rice monitoring. Synthetic aperture radar is not affected by weather and can obtain backscatter information below the rice canopy, which is beneficial to long-term continuous monitoring of rice. Time series SAR images have the ability to observe the entire phenological cycle of rice and capture rice growth status and phenological changes, and can provide discriminative remote sensing features for rice identification. However, most of the existing research focuses on rice planting areas with a single rice planting pattern, or usually studies different rice planting patterns (single-cropping rice, double-cropping rice) separately, and less considers the integrated identification research of rice fields under complex planting patterns, making it difficult to Reflecting the real rice planting situation under various planting modes, the understanding of the spatial distribution of overall rice planting in a region is limited.

Contents of the invention

The purpose of the present invention is to propose an integrated remote sensing identification method for multi-planting model rice fields in view of the shortcomings of the existing technology.

The first aspect provides an integrated remote sensing identification method for rice fields with multiple planting patterns, including:

Step 1. Collect digital elevation model data of the study area and perform preprocessing to obtain elevation and slope image data of the study area;

Step 2. Collect the SAR remote sensing image data throughout the year and preprocess it, convert the SAR remote sensing image pixel values into actual surface object backscattering intensity, and obtain the annual time series surface object backscattering intensity data;

Step 3. Collect sample points of rice and non-rice categories as training samples; the rice includes single-crop early rice, single-crop mid-rice, single-crop late rice, and double-crop rice; the non-rice includes buildings, natural vegetation, water bodies, and upland farming. crop;

Step 4. Obtain rice phenology data, which includes the dates from sowing to maturity and harvesting of rice;

Step 5: Calculate and extract features based on the backscattering intensity data of ground objects in the time series throughout the year. The features include statistical feature data and rice phenological parameter feature data;

Step 6: Overlay the SAR backscattering intensity data, elevation and slope image data, statistical feature data and rice phenological parameter feature data to obtain combined multi-band feature image data;

Step 7. Based on the multi-band characteristic image data and sample points of rice and non-rice categories, use a random forest classifier to identify rice.

Preferably, in step 1, the digital elevation model data is elevation image data with a resolution of no less than 30 meters.

Preferably, in step 2, the pretreatment includes:

Step 2.1. Perform radiation calibration on SAR remote sensing image data. The formula is:

Among them, σ ₀ is the backscattering coefficient, A _i is the backscatter returned to the antenna from pixel i per unit time, and DN _i is the gray value of pixel i;

Step 2.2: Use mean filtering to filter the SAR remote sensing image data; the convolution kernel size used in mean filtering is 3×3, the central element value is 1, the surrounding element value is 1, and the convolution kernel is:

Step 2.3. Use elevation data to perform Doppler terrain correction on SAR remote sensing image data;

Step 2.4. Convert the SAR remote sensing image data to decibels and convert the unitless backscattering intensity into decibels. The formula is:

σ＝10*log ₁₀ σ ₀

Among them, σ ₀ is the original backscattering coefficient, and σ represents the backscattering coefficient after decibel conversion, in dB;

Step 2.5: Perform SG filtering on the decibelized backscattering intensity data of ground objects throughout the year.

As a preference, in step 2.5, the calculation formula is:

In the above formula, Y _j ^* is the j-th reconstruction value; C _i is the coefficient of the i-th point in the sliding window; N is the length of the sliding window, its size is equal to 2m+1, m is the length of half the sliding window, m Set to 3-7.

Preferably, in step 3, there should be no less than 100 sample points for single-crop early rice, single-crop mid-rice, single-crop late rice, double-crop rice, buildings, natural vegetation, water bodies and upland crops respectively.

Preferably, step 5 includes:

Step 5.1. Extract statistical features, which include mean, median, range, sum of squares of deviations, standard deviation and coefficient of variation;

Step 5.2. Extract the phenological characteristics of rice. The phenological characteristics of rice include the transplanting date T _TD of early rice, middle rice and late rice, the maturity date T _MD , the length of the growing season GSL, the change rate of backscattering coefficient from sowing to transplanting period V _ST and The change rate V _TM of the backscattering coefficient from transplanting to the mature stage.

As a preference, in step 5.2, assume that the time series backscattering coefficient and corresponding time of a pixel in the early rice phenological period are expressed as (d ₁ ,σ ₁ ), (d ₂ ,σ ₂ ),..., (d _n , σ _n ), where σ ₁ ,…,σ _n are the year-round time series backscattering coefficients of the pixel, and d ₁ ,…,d _n are the dates corresponding to σ ₁ ,…,σ _n ;

If σ _i =min(σ ₁ ,…,σ _n ), then the transplanting date T _TD is _di ; where the transplanting date T _TD represents the date corresponding to the minimum value of the pixel backscatter coefficient time series curve in one year date sequence;

If σ _j =max(σ ₁ ,…,σ _n ), then the maturity date T _MD is d _j ;

The growing season length GSL=d _j -d _i ; where the growing season length GSL represents the time difference between the maturity date and the transplanting date.

Assuming that (d _i1 , σ _i1 ) is the rice sowing date and the corresponding backscattering coefficient, (d _i2 , σ _i2 ) is the rice transplanting date and the corresponding backscattering coefficient, then the backscattering coefficient from sowing to transplanting period Change rate V _ST =(σ _i2 -σ _i1 )/(d _i2 -d _i1 );

Assuming that (d _i3 ,σ _i3 ) is the maturity date of rice and the corresponding backscattering coefficient, then the change rate of backscattering coefficient from transplanting to the maturity stage V _TM = (σ _i3 -σ _i3 )/(d _i2 -d _i2 ) .

As a preference, the number of decision trees in the random forest algorithm in step 7 is 50-150.

In the second aspect, an integrated remote sensing identification system for rice fields with multiple planting patterns is provided, which is used to implement the integrated remote sensing identification method for rice fields with multiple planting patterns described in any one of the first aspects, including:

The first collection module is used to collect digital elevation model data of the study area and perform preprocessing to obtain elevation and slope image data of the study area;

The second collection module is used to collect SAR remote sensing image data throughout the year and preprocess it, convert the SAR remote sensing image pixel values into actual backscattering intensity of ground objects, and obtain the backscattering intensity of ground objects in the time series throughout the year. data;

The collection module is used to collect sample points of rice and non-rice categories as training samples; the rice includes single-season early rice, single-season mid-rice, single-season late rice and double-season rice; the non-rice includes buildings, natural vegetation, water bodies and rainfed crops;

An acquisition module, used to acquire rice phenology data, the rice phenology data including the dates from sowing to maturity and harvesting of rice;

The calculation module is used to calculate the backscatter intensity data of ground objects in time series throughout the year and extract features, where the features include statistical feature data and rice phenological parameter feature data;

The overlay module is used to superimpose SAR backscattering intensity data, elevation and slope image data, statistical feature data and rice phenological parameter feature data to obtain combined multi-band feature image data;

The identification module is used to identify rice using a random forest classifier based on multi-band characteristic image data, sample points of rice and non-rice categories.

In a third aspect, a computer-readable storage medium is provided. The computer-readable storage medium includes a stored executable program, wherein when the executable program is running, the device where the computer-readable storage medium is located is controlled to execute the first step. On the one hand, any one of the above described integrated remote sensing identification methods for rice fields with multiple planting patterns.

The beneficial effects of the present invention are: by using SAR images, the present invention avoids the impact of weather conditions on image imaging quality and subsequent rice identification and monitoring; and, the present invention realizes the distinction between rice and non-rice through statistical characteristics, and on this basis Through phenological characteristics, the distinction between single-crop early rice, single-crop mid-season rice, single-crop late rice and double-crop rice under various planting modes was achieved, and more accurate and precise remote sensing monitoring results of rice fields were obtained.

Description of the drawings

Figure 1 is a technical flow chart of an integrated remote sensing identification method for rice fields with multiple planting patterns;

Figure 2 is a comparison chart of the statistical characteristics of rice and non-rice types in the embodiment of the present invention;

Figure 3 is a histogram of phenological characteristic parameters of rice and non-rice in the embodiment of the present invention;

Figure 4 is a diagram showing the identification results of single-crop early rice, single-crop mid-rice, single-crop late rice and double-crop rice in the embodiment of the present invention.

Detailed ways

The present invention will be further described below in conjunction with examples. The following description of the examples is provided only to assist understanding of the present invention. It should be pointed out that for those skilled in the art, several modifications can be made to the present invention without departing from the principle of the present invention, and these improvements and modifications also fall within the protection scope of the claims of the present invention.

Example 1:

Embodiment 1 of the present application provides an integrated remote sensing identification method for rice fields with multiple planting patterns, as shown in Figure 1, including:

Step 1. Collect digital elevation model data of the study area and perform preprocessing to obtain elevation and slope image data of the study area;

Step 2. Collect the SAR remote sensing image data throughout the year and preprocess it, convert the SAR remote sensing image pixel values into actual surface object backscattering intensity, and obtain the annual time series surface object backscattering intensity data;

Step 3. Collect sample points of rice and non-rice categories as training samples; the rice includes single-crop early rice, single-crop mid-rice, single-crop late rice, and double-crop rice; the non-rice includes buildings, natural vegetation, water bodies, and upland farming. crop;

Step 4. Obtain rice phenology data, which includes the dates from sowing to maturity and harvesting of rice;

Step 5: Calculate and extract features based on the backscattering intensity data of ground objects in the time series throughout the year. The features include statistical feature data and rice phenological parameter feature data;

Step 6: Overlay the SAR backscattering intensity data, elevation and slope image data, statistical feature data and rice phenological parameter feature data to obtain combined multi-band feature image data;

Step 7. Based on the multi-band characteristic image data and sample points of rice and non-rice categories, use a random forest classifier to identify rice.

Example 2:

On the basis of Example 1, Example 2 of the present application provides the practical application of an integrated remote sensing identification method for rice fields with multiple planting patterns in Example 1: Ningbo City, Zhejiang Province is a typical rice field with multiple planting patterns. Area, single-season early rice, single-season mid-season rice, single-season late rice and double-season rice are planted in this area, with diverse rice planting patterns and complex planting structures; an integrated remote sensing identification method for multi-planting pattern rice fields is applied to Zhejiang Province Ningbo City; Use Sentinel-1 images to monitor the planting conditions of single-crop early rice, single-crop mid-season rice, single-crop late rice and double-crop rice. As shown in Figure 1, the method of this embodiment includes the following steps:

Step 1. Collect digital elevation model data of the study area and perform preprocessing to obtain elevation and slope image data of the study area.

In step 1, the digital elevation model data is elevation image data with a resolution of no less than 30 meters. In addition, the nearest neighbor method can be used to resample the elevation data to 10 meter resolution.

Step 2: Collect SAR remote sensing image data throughout the year and preprocess it, convert the SAR remote sensing image pixel values into actual surface object backscatter intensity, and obtain the annual time series surface object backscattering intensity data.

Step 2 includes:

Step 2.1. Perform radiation calibration on SAR remote sensing image data:

In the above formula, σ ₀ is the backscattering coefficient, A _i is the backscatter returned to the antenna from pixel i per unit time, and DN _i is the gray value of pixel i;

Step 2.2: Use mean filtering to filter the SAR remote sensing image data. The convolution kernel size used in mean filtering is 3×3, the central element value is 1, the surrounding element value is 1, and the convolution kernel is:

Step 2.3. Use elevation data to perform Doppler terrain correction on SAR remote sensing image data;

Step 2.4. Convert the SAR remote sensing image data to decibels and convert the unitless (linear) backscattering intensity into decibels (dB):

σ＝10*log ₁₀ σ ₀

In the above formula, σ ₀ is the original backscattering coefficient, and σ represents the backscattering coefficient after decibel conversion, in dB.

Step 2.5: Perform Savitzky-Golay filtering on the decibelized annual time series backscattering coefficient data:

In the above formula, Y _j ^* is the j-th reconstruction value; C _i is the coefficient of the i-th point in the sliding window; N is the length of the sliding window, its size is equal to 2m+1, m is the length of half the sliding window, m Set to 5.

Step 3. Collect sample points of rice and non-rice categories as training samples; the rice includes single-crop early rice, single-crop mid-rice, single-crop late rice, and double-crop rice; the non-rice includes buildings, natural vegetation, water bodies, and upland farming. crop.

Specifically, rice and non-rice sample points were selected based on field survey data, drone images and high-resolution Google Earth images, including single-crop early rice, single-crop mid-rice, single-crop late rice, double-crop rice, buildings, natural vegetation, water bodies The number of sample points for agricultural and dry crops shall not be less than 100 respectively.

Step 4: Obtain rice phenology data, which includes the dates from sowing to maturity and harvesting of rice.

As an example, rice phenology data are obtained through field observations.

Step 5: Calculate and extract features based on the backscattering intensity data of ground objects in the time series throughout the year. The features include statistical feature data and rice phenological parameter feature data.

Step 5 includes:

Step 5.1. As shown in Figure 2, extract statistical features. The statistical features include mean, median, range, sum of squares of deviations, standard deviation and coefficient of variation; the specific formula is as follows:

Mean:

Median value: median=σ _(n+1)/2 , n is an odd number; median=σ _n/2 , n is an even number

Range: range=σ _max -σ _min

Sum of squared deviations:

Standard deviation:

Coefficient of variation:

In the formula, σ is the backscattering coefficient, and n is the number of images.

Step 5.2, as shown in Figure 3, extract rice phenological characteristics, which include transplanting date _TTD , maturity date _TMD , growing season length GSL, and backscatter from sowing to transplanting period of early rice, middle rice, and late rice. The coefficient change rate V _ST and the backscattering coefficient change rate V _TM after transplanting to the maturity stage.

In step 5.2, it is assumed that the time series backscattering coefficient and corresponding time of a pixel during the early rice phenological period are expressed as (d ₁ ,σ ₁ ), (d ₂ ,σ ₂ ),…, (d _n ,σ _n ) , where σ ₁ ,…,σ _n is the backscattering coefficient of the pixel’s time series throughout the year, and d ₁ ,…,d _n is the date corresponding to σ ₁ ,…,σ _n ;

If σ _i =min(σ ₁ ,…,σ _n ), then the transplanting date T _TD is _di ; where the transplanting date T _TD represents the date corresponding to the minimum value of the pixel backscatter coefficient time series curve in one year date sequence;

If σ _j =max(σ ₁ ,…,σ _n ), then the maturity date T _MD is d _j ;

The growing season length GSL=d _j -d _i ; where the growing season length GSL represents the time difference between the maturity date and the transplanting date.

Assuming that (d _i1 , σ _i1 ) is the rice sowing date and the corresponding backscattering coefficient, (d _i2 , σ _i2 ) is the rice transplanting date and the corresponding backscattering coefficient, then the backscattering coefficient from sowing to transplanting period Change rate V _ST =(σ _i2 -σ _i1 )/(d _i2 -d _i1 );

Assuming that (d _i3 ,σ _i3 ) is the maturity date of rice and the corresponding backscattering coefficient, then the change rate of backscattering coefficient from transplanting to the maturity stage V _TM = (σ _i3 -σ _i3 )/(d _i2 -d _i2 ) .

Step 6: Overlay the SAR backscattering intensity data, elevation and slope image data, statistical feature data and rice phenological parameter feature data to obtain combined multi-band feature image data.

Step 7. As shown in Figure 4, based on multi-band characteristic image data, rice (single-crop early rice, single-crop mid-rice, single-crop late rice and double-crop rice) and non-rice (buildings, natural vegetation, water bodies, upland crops) categories At the sample points, a random forest classifier is used to identify rice.

Specifically, the number of decision trees in the random forest algorithm in step 7 is 50-150, and the optimal number is 100.

It should be noted that the same or similar parts in this embodiment as those in Embodiment 1 may be referred to each other and will not be described again in this application.

Example 3:

On the basis of Embodiments 1 and 2, Embodiment 3 of the present application provides an integrated remote sensing identification system for rice fields with multiple planting patterns, including:

The first collection module is used to collect digital elevation model data of the study area and perform preprocessing to obtain elevation and slope image data of the study area;

The second collection module is used to collect SAR remote sensing image data throughout the year and preprocess it, convert the SAR remote sensing image pixel values into actual backscattering intensity of ground objects, and obtain the backscattering intensity of ground objects in the time series throughout the year. data;

The collection module is used to collect sample points of rice and non-rice categories as training samples; the rice includes single-season early rice, single-season mid-rice, single-season late rice and double-season rice; the non-rice includes buildings, natural vegetation, water bodies and rainfed crops;

An acquisition module, used to acquire rice phenology data, the rice phenology data including the dates from sowing to maturity and harvesting of rice;

A calculation module, used to calculate and extract features based on the backscattering intensity data of ground objects in time series throughout the year, where the features include statistical feature data and rice phenological parameter feature data;

The overlay module is used to superimpose SAR backscattering intensity data, elevation and slope image data, statistical feature data and rice phenological parameter feature data to obtain combined multi-band feature image data;

The identification module is used to identify rice using a random forest classifier based on multi-band characteristic image data, sample points of rice and non-rice categories.

Specifically, the system provided in this embodiment is a system corresponding to the method provided in Embodiment 1. Therefore, the same or similar parts in this embodiment as those in Embodiment 1 can be referred to each other and will not be described again in this application.

Claims (10)

1. An integrated remote sensing identification method for rice fields with multiple planting patterns, which is characterized by including: Step 1. Collect digital elevation model data of the study area and perform preprocessing to obtain elevation and slope image data of the study area; Step 2. Collect the SAR remote sensing image data throughout the year and preprocess it, convert the SAR remote sensing image pixel values into actual surface object backscattering intensity, and obtain the annual time series surface object backscattering intensity data; Step 3. Collect sample points of rice and non-rice categories as training samples; the rice includes single-crop early rice, single-crop mid-rice, single-crop late rice, and double-crop rice; the non-rice includes buildings, natural vegetation, water bodies, and upland farming. crop; Step 4. Obtain rice phenology data, which includes the dates from sowing to maturity and harvesting of rice; Step 5: Calculate and extract features based on the backscattering intensity data of ground objects in the time series throughout the year. The features include statistical feature data and rice phenological parameter feature data; Step 6: Overlay the SAR backscattering intensity data, elevation and slope image data, statistical feature data and rice phenological parameter feature data to obtain combined multi-band feature image data; Step 7. Based on the multi-band characteristic image data and sample points of rice and non-rice categories, use a random forest classifier to identify rice. 2. The integrated remote sensing identification method for rice fields with multiple planting patterns according to claim 1, characterized in that in step 1, the digital elevation model data is elevation image data with a resolution of not less than 30 meters. 3. The integrated remote sensing identification method for rice fields with multiple planting patterns according to claim 2, characterized in that in step 2, the preprocessing includes: Step 2.1. Perform radiation calibration on SAR remote sensing image data. The formula is: Among them, σ ₀ is the backscattering coefficient, A _i is the backscatter returned to the antenna from pixel i per unit time, and DN _i is the gray value of pixel i; Step 2.2: Use mean filtering to filter the SAR remote sensing image data. The convolution kernel size used in mean filtering is 3×3, the central element value is 1, the surrounding element value is 1, and the convolution kernel is: Step 2.3. Use ALOS elevation data to perform Doppler terrain correction on SAR remote sensing image data; Step 2.4. Convert the SAR remote sensing image data to decibels and convert the unitless backscattering intensity into decibels. The formula is: σ＝10*log ₁₀ σ ₀ Among them, σ ₀ is the original backscattering coefficient, and σ represents the backscattering coefficient after decibel conversion, in dB; Step 2.5: Perform SG filtering on the decibelized backscattering intensity data of ground objects throughout the year. 4. The integrated remote sensing identification method for multi-planting model rice fields according to claim 3, characterized in that in step 2.5, the calculation formula is: In the above formula, Y _j ^* is the j-th reconstruction value; C _i is the coefficient of the i-th point in the sliding window; N is the length of the sliding window, its size is equal to 2m+1, m is the length of half the sliding window, m Set to 3-7. 5. The integrated remote sensing identification method for multi-planting model rice fields according to claim 4, characterized in that in step 3, single-season early rice, single-season mid-rice, single-season late rice, double-season rice, buildings, and natural vegetation There should be no less than 100 sample points for water bodies and upland crops respectively. 6. The integrated remote sensing identification method for multi-planting model rice fields according to claim 5, characterized in that step 5 includes: Step 5.1. Extract statistical features, which include mean, median, range, sum of squares of deviations, standard deviation and coefficient of variation; Step 5.2. Extract the phenological characteristics of rice. The phenological characteristics of rice include the transplanting date T _TD of early rice, middle rice and late rice, the maturity date T _MD , the length of the growing season CSL, the change rate of backscattering coefficient from sowing to transplanting period V _ST and The change rate V _TM of the backscattering coefficient from transplanting to the mature stage. 7. The integrated remote sensing identification method for multi-planting model rice fields according to claim 6, characterized in that in step 5.2, it is assumed that the time series backscattering coefficient and corresponding time representation of a pixel in the early rice phenological period are (d ₁ ,σ ₁ ), (d ₂ ,σ ₂ ),…, (d _n ,σ _n ), where σ ₁ ,…,σ _n is the backscattering coefficient of the pixel’s time series throughout the year, i ₁ ,…,d _n is the date corresponding to σ ₁ ,…,σ _n ; If σ _i =min(σ ₁ ,…,σ _n ), then the transplanting date T _TD is _di ; where the transplanting date T _TD represents the date corresponding to the minimum value of the pixel backscatter coefficient time series curve in one year date sequence; If σ _j =max(σ ₁ ,…,σ _n ), then the maturity date T _MD is d _j ; The growing season length GSL=d _j -d _i ; where the growing season length GSL represents the time difference between the maturity date and the transplanting date. Assuming that (d _i1 , σ _i1 ) is the rice sowing date and the corresponding backscattering coefficient, (d _i2 , σ _i2 ) is the rice transplanting date and the corresponding backscattering coefficient, then the backscattering coefficient from sowing to transplanting period Change rate V _ST =(σ _i2 -σ _i1 )/(d _i2 -d _i1 ); Assuming that (d _i3 ,σ _i3 ) is the maturity date of rice and the corresponding backscattering coefficient, then the change rate of backscattering coefficient from transplanting to the maturity stage V _TM = (σ _i3 -σ _i3 )/(d _i2 -d _i2 ) . 8. The integrated remote sensing identification method for rice fields with multiple planting patterns according to claim 6, characterized in that the number of decision trees in the random forest algorithm in step 7 is 50-150. 9. An integrated remote sensing identification system for rice fields with multiple planting patterns, characterized in that it is used to perform the integrated remote sensing identification method for rice fields with multiple planting patterns described in any one of claims 1 to 8, including: The first collection module is used to collect digital elevation model data of the study area and perform preprocessing to obtain elevation and slope image data of the study area; The second collection module is used to collect SAR remote sensing image data throughout the year and preprocess it, convert the SAR remote sensing image pixel values into actual backscattering intensity of ground objects, and obtain the backscattering intensity of ground objects in the time series throughout the year. data; The collection module is used to collect sample points of rice and non-rice categories as training samples; the rice includes single-season early rice, single-season mid-rice, single-season late rice and double-season rice; the non-rice includes buildings, natural vegetation, water bodies and rainfed crops; An acquisition module, used to acquire rice phenology data, the rice phenology data including the dates from sowing to maturity and harvesting of rice; A calculation module, used to calculate and extract features based on the backscattering intensity data of ground objects in time series throughout the year, where the features include statistical feature data and rice phenological parameter feature data; The overlay module is used to superimpose SAR backscattering intensity data, elevation and slope image data, statistical feature data and rice phenological parameter feature data to obtain combined multi-band feature image data; The identification module is used to identify rice using a random forest classifier based on multi-band characteristic image data, sample points of rice and non-rice categories. 10. A computer-readable storage medium, characterized in that the computer-readable storage medium includes a stored executable program, wherein when the executable program is running, the device where the computer-readable storage medium is located is controlled to execute the right The integrated remote sensing identification method for rice fields with multiple planting patterns described in any one of claims 1 to 8.