High-spectrum image segmentation method based on pixel space information

Abstract

The invention discloses a high-spectrum image segmentation method based on pixel space information, mainly solving the problem that similar physiognomies can not be favorably segmented by the prior method. The high-spectrum image segmentation method comprises the following steps: solving and normalizing a pixel characteristics matrix and a pixel space Euclidean distance matrix of high-spectrum data; weighting the pixel characteristics matrix and the pixel space Euclidean distance matrix, adding the two weighted matrixes to form a joint dissimilarity matrix and adjusting weighted parameters toacquire a plurality of groups of joint dissimilarity matrixes; using an isometric mapping algorithm to reduce the dimension of each group of joint dissimilarity matrix and acquiring a plurality of groups of mapping results; counting and analyzing each group of mapping result, finishing the primary segmentation of a high-spectrum image; and carrying out category correction to primarily segmented boundary points to acquire a final image segmentation result. The method can effectively find the nuance of different physiognomies in the high-spectrum image and can be applied to martial object recognition, mineral exploration and environmental condition analysis.

Description

Technical field

[0001] The invention belongs to the technical field of image processing, in particular to an image segmentation method, which can analyze complex landforms, find interesting targets from similar landforms and roughly delimit the boundaries between the landforms. Background technique

[0002] Image segmentation refers to processing natural images or SAR images to separate different targets in the image. Hyperspectral image data is generated by hyperspectral remote sensing and has a high dimensionality. Each dimension of its data is a grayscale image. The traditional image segmentation method cannot be used to process it as a whole. It is too much work to process each dimension of the data and then combine it, and the effect is difficult to guarantee. Some foreign scholars have proposed the use of support vector machines to segment hyperspectral images, and the effect is also good. However, the support vector machine is a supervised learning method. It is necessar...

Examples

Embodiment Construction

[0021] Reference figure 1 The specific implementation steps of the present invention are as follows:

[0022] Step 1: Solve the Euclidean distance between the pixel feature information of the hyperspectral image data and the pixel spatial information, and normalize it.

[0023] 1.1) Select the hyperspectral image image 3 In the area shown, the data of this area contains 200 bands, and each band is a gray-scale image. Therefore, these data can be regarded as 200 gray-scale feature matrices, and each gray-scale feature matrix is ​​divided into band order Decompose according to the columns, and then connect them in order to form a large column vector, thereby obtaining 200 column vectors;

[0024] 1.2) Combine these 200 column vectors to form a pixel feature matrix data, each row of the feature matrix data represents the pixel feature vector of a pixel, and calculate the Euclidean distance of each two rows of pixel feature vectors to obtain the pixel feature Euclidean Distance matrix ...

Technical field

[0001] The invention belongs to the technical field of image processing, in particular to an image segmentation method, which can analyze complex landforms, find interesting targets from similar landforms and roughly delimit the boundaries between the landforms. Background technique

[0002] Image segmentation refers to processing natural images or SAR images to separate different targets in the image. Hyperspectral image data is generated by hyperspectral remote sensing and has a high dimensionality. Each dimension of its data is a grayscale image. The traditional image segmentation method cannot be used to process it as a whole. It is too much work to process each dimension of the data and then combine it, and the effect is difficult to guarantee. Some foreign scholars have proposed the use of support vector machines to segment hyperspectral images, and the effect is also good. However, the support vector machine is a supervised learning method. It is necessar...

Examples

Embodiment Construction

[0021] Reference figure 1 The specific implementation steps of the present invention are as follows:

[0022] Step 1: Solve the Euclidean distance between the pixel feature information of the hyperspectral image data and the pixel spatial information, and normalize it.

[0023] 1.1) Select the hyperspectral image image 3 In the area shown, the data of this area contains 200 bands, and each band is a gray-scale image. Therefore, these data can be regarded as 200 gray-scale feature matrices, and each gray-scale feature matrix is ​​divided into band order Decompose according to the columns, and then connect them in order to form a large column vector, thereby obtaining 200 column vectors;

[0024] 1.2) Combine these 200 column vectors to form a pixel feature matrix data, each row of the feature matrix data represents the pixel feature vector of a pixel, and calculate the Euclidean distance of each two rows of pixel feature vectors to obtain the pixel feature Euclidean Distance matrix ...

Images