CN106530326A - Change detection method based on image texture features and DSM - Google Patents

Abstract

The present invention provides a change detection method based on image texture features and DSM, comprising the following steps: Step 1, dividing ground objects into multiple types of ground objects; according to the characteristics of each type of ground objects, setting the corresponding At the same time, the statistical analysis obtains the corresponding texture features of each surface object; for each sub-area, according to the type of surface objects corresponding to the sub-area, the corresponding texture features and corresponding texture features set in step 1 are retrieved. Elevation change threshold range: use the texture feature and elevation change threshold range corresponding to the feature type to further judge the preliminary change detection result, and determine whether the preliminary change detection result is correct. The method has the following advantages: the present invention integrates texture features and DSM information into the change detection process in the manner of different ground object thresholds, thereby effectively improving the change detection accuracy.

Description

A Change Detection Method Based on Image Texture Features and DSM

technical field

The invention belongs to the technical field of change detection, in particular to a change detection method based on image texture features and DSM.

Background technique

Remote sensing image change detection technology refers to: by analyzing two or more images from the same area at different times, it detects the change information of the ground features in the area over time. At present, change detection technology has developed rapidly and is widely used in environmental monitoring, land use, crop growth monitoring, disaster estimation and other fields.

According to the level of processing information, change detection technology is divided into three levels: pixel level, feature level and decision level. With the complexity of land cover changes and the increase in the diversity of remote sensing data, new change detection methods and new image processing algorithms Constructing a stable and reliable change detection algorithm has always been a key research direction in the field of change detection.

The existing change detection technology only selects a single image feature in two-dimensional space as the processing object and the basis for discrimination. This change detection method mainly has the following disadvantages: the change detection method has very strict requirements on image data. However, the quality of images in different imaging conditions and different time phases is often uneven. Therefore, the traditional change detection method has the problem of limited change detection accuracy, and it is difficult to meet the needs of high detection accuracy in specific fields.

Contents of the invention

Aiming at the defects in the prior art, the present invention provides a change detection method based on image texture features and DSM, which can effectively solve the above problems.

The technical scheme that the present invention adopts is as follows:

The present invention provides a kind of change detection method based on image texture feature and DSM, comprises the following steps:

Step 1. Divide the ground features into multiple types of ground features; according to the characteristics of each ground feature type, set the elevation change threshold range corresponding to each ground feature; at the same time, obtain the texture characteristics corresponding to each ground feature through statistical analysis;

Step 2, the two remote sensing images acquired at different times in the same region are referred to as the first original image and the second original image; after preprocessing the first original image and the second original image respectively, the first 1 performing initial change detection on the original image and the second original image, determining a preliminary change area of the first original image relative to the second original image, and obtaining a preliminary change detection result;

Step 3, segmenting the preliminary change area according to the type of feature, and dividing it into sub-areas containing only one type of feature;

Step 4, for each sub-area, according to the feature type corresponding to the sub-area, call the corresponding texture feature and the corresponding elevation change threshold range set in step 1; use the texture feature corresponding to the feature type and The elevation change threshold range is further judged on the preliminary change detection result to determine whether the preliminary change detection result is correct.

Preferably, in step 1, the elevation change threshold range is obtained by the following method:

A detection sensor is mounted on a ground platform, an aerospace platform or an aviation platform; through the detection sensor, an elevation change threshold range corresponding to each ground object is obtained.

Preferably, the ground platform is equipped with the following detection sensors: a three-dimensional laser scanner or a digital camera;

The space platform is equipped with the following detection sensors: one of spaceborne three-dimensional lidar, spaceborne synthetic aperture radar and spaceborne multispectral sensor;

The aerial platform is equipped with the following detection sensors: image data detection sensor, 3D laser point cloud data detection sensor and radar data detection sensor.

Preferably, the texture features corresponding to each ground object are obtained through statistical texture analysis or structural texture analysis.

Preferably, the texture features obtained through statistical texture analysis are digital features of statistical textures, including one or more of autocorrelation function, gray level co-occurrence matrix, gray level run and gray level distribution of local image regions.

Preferably, the texture features obtained by structural texture analysis include: one of energy, contrast, correlation, entropy, inverse gap, median, covariance, homogeneity, contrast, heterogeneity, second-order distance, and autocorrelation or several.

The change detection method based on image texture features and DSM provided by the present invention has the following advantages:

The invention integrates the texture features and DSM information into the change detection process in the manner of different ground object thresholds, thereby effectively improving the change detection accuracy.

Description of drawings

FIG. 1 is a schematic flowchart of a change detection method based on image texture features and DSM provided by the present invention.

detailed description

In order to make the technical problems, technical solutions and beneficial effects solved by the present invention clearer, the present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present invention, not to limit the present invention.

In order to weaken the impact of a single feature index and image quality on change detection, the present invention provides a change detection method based on image texture features and DSM (Digital Surface Model, DSM), which extends image change detection from two-dimensional space to three-dimensional space, Improve change detection accuracy. The change detection method based on image texture features and DSM provided by the present invention comprises the following steps:

Step 1. Divide the ground features into multiple types of ground features; according to the characteristics of each ground feature type, set the elevation change threshold range corresponding to each ground feature; at the same time, obtain the texture characteristics corresponding to each ground feature through statistical analysis;

In this step, the elevation change threshold range is obtained by the following method: a ground platform, space platform or aviation platform is equipped with a detection sensor; through the detection sensor, the elevation change threshold range corresponding to each ground object is obtained.

Specifically, the ground platform is equipped with the following detection sensors: a three-dimensional laser scanner or a digital camera (which can form a stereo image pair).

The space platform is equipped with the following detection sensors: one of spaceborne three-dimensional laser radar (LIDAR), spaceborne synthetic aperture radar (INSAR) and spaceborne multispectral sensors (which can form stereo image pairs);

The aerial platform is equipped with the following detection sensors: image data detection sensor, 3D laser point cloud data detection sensor and radar data detection sensor. Specifically include: aerial digital camera, high-pixel aerial digital camera, RC30 aerial camera, RMKTOP aerial camera, LMK2000 aerial camera, DMC full digital aerial camera, Nikon camera, Canon camera, 3D laser radar (LIDAR), low-altitude digital remote sensing One or more of the airborne synthetic aperture radar (INSAR), airborne imaging spectrometer.

The texture features corresponding to each surface object are obtained through statistical texture analysis or structural texture analysis. Among them, the texture feature obtained by statistical texture analysis is the digital feature of statistical texture, including one or more of autocorrelation function, gray level co-occurrence matrix, gray level run and gray level distribution in the local area of the image. Texture features obtained through structural texture analysis include: one or more of energy, contrast, correlation, entropy, inverse gap, median, covariance, homogeneity, contrast, heterogeneity, second-order distance, and autocorrelation .

Step 2, the two remote sensing images acquired at different times in the same region are referred to as the first original image and the second original image; after preprocessing the first original image and the second original image respectively, the first 1 performing initial change detection on the original image and the second original image, determining a preliminary change area of the first original image relative to the second original image, and obtaining a preliminary change detection result;

In terms of specific implementation, the preliminary change detection results are obtained through the following methods:

The original images of the two time phases are subjected to image preprocessing, including radiometric calibration, atmospheric correction, geometric correction, image fusion, image registration, and image texture feature extraction to obtain preprocessed remote sensing images; The DSM data of phase, the DSM data collection time and the satellite image collection time must keep consistent.

The two time-phase images select common samples (that is, select appropriate samples in the unchanged area), add image feature parameters, and perform image supervision classification. The algorithm can use maximum likelihood method, Mahalanobis distance method, etc. to obtain classification results.

Based on the classification results of the two-temporal data, the difference or transformation algorithm is used for classification and transformation detection, and the preliminary results are obtained. At the same time, the change attributes and directions can also be obtained.

Step 3, segmenting the preliminary change area according to the type of feature, and dividing it into sub-areas containing only one type of feature;

Step 4, for each sub-area, according to the feature type corresponding to the sub-area, call the corresponding texture feature and the corresponding elevation change threshold range set in step 1; use the texture feature corresponding to the feature type and The elevation change threshold range is further judged on the preliminary change detection result to determine whether the preliminary change detection result is correct.

In the present invention, by setting the elevation change threshold ranges corresponding to different ground objects, for example, when the forest information changes, a range of elevation change thresholds is drawn up, which conforms to the law of the average height change of local trees; and when the building land changes, it also Develop a threshold range for elevation change that is known to be significantly different from the threshold range for trees.

The change detection of elevation information is added to increase the judgment conditions of the change area, which greatly reduces the misjudgment of change information caused by the spectral defect of "same object with different spectrum, and different object with same spectrum". A more accurate orthophoto will reduce errors caused by shadow areas of ground objects or poor projection.

It can be seen that the change detection method based on image texture features and DSM provided by the present invention has the following advantages:

The invention integrates the texture features and DSM information into the change detection process in the manner of different ground object thresholds, thereby effectively improving the change detection accuracy.

The above is only a preferred embodiment of the present invention, and it should be pointed out that for those of ordinary skill in the art, some improvements and modifications can also be made without departing from the principles of the present invention. It should be regarded as the protection scope of the present invention.

Claims (6)

1. A change detection method based on image texture features and DSM, characterized in that, comprising the following steps: Step 1. Divide the ground features into multiple types of ground features; according to the characteristics of each ground feature type, set the elevation change threshold range corresponding to each ground feature; at the same time, obtain the texture characteristics corresponding to each ground feature through statistical analysis; Step 2, the two remote sensing images acquired at different times in the same region are referred to as the first original image and the second original image; after preprocessing the first original image and the second original image respectively, the first 1 performing initial change detection on the original image and the second original image, determining a preliminary change area of the first original image relative to the second original image, and obtaining a preliminary change detection result; Step 3, segmenting the preliminary change area according to the type of feature, and dividing it into sub-areas containing only one type of feature; Step 4, for each sub-area, according to the feature type corresponding to the sub-area, call the corresponding texture feature and the corresponding elevation change threshold range set in step 1; use the texture feature corresponding to the feature type and The elevation change threshold range is further judged on the preliminary change detection result to determine whether the preliminary change detection result is correct. 2. the change detection method based on image texture feature and DSM according to claim 1, is characterized in that, in step 1, described height change threshold range obtains by following method: A detection sensor is mounted on a ground platform, an aerospace platform or an aviation platform; through the detection sensor, an elevation change threshold range corresponding to each ground object is obtained. 3. The change detection method based on image texture features and DSM according to claim 2, wherein the following detection sensors are carried on the ground platform: a three-dimensional laser scanner or a digital camera; The space platform is equipped with the following detection sensors: one of spaceborne three-dimensional lidar, spaceborne synthetic aperture radar and spaceborne multispectral sensor; The aerial platform is equipped with the following detection sensors: image data detection sensor, 3D laser point cloud data detection sensor and radar data detection sensor. 4. The change detection method based on image texture features and DSM according to claim 1, wherein the texture features corresponding to each kind of ground object are obtained through statistical texture analysis or structural texture analysis. 5. the change detection method based on image texture feature and DSM according to claim 4, it is characterized in that, the texture feature that obtains by statistical texture analysis is the digital feature of statistical texture, comprises the autocorrelation function of image local area, gray level One or more of co-occurrence matrix, gray run and gray distribution. 6. The change detection method based on image texture features and DSM according to claim 4, wherein the texture features obtained by structural texture analysis include: energy, contrast, correlation, entropy, inverse gap, median, covariance , homogeneity, contrast, heterogeneity, second-order distance, and autocorrelation.