Image texture feature extraction method based on nested triangular structure

A triangular structure and image texture technology, applied in image data processing, image analysis, instruments, etc., can solve the problems of incomplete texture image information, missing image information, etc., to reduce the amount of calculation of statistical features, less sampling points, and good stability sexual effect

Active Publication Date: 2017-08-18
5 Cites 0 Cited by

AI-Extracted Technical Summary

Problems solved by technology

For example, GLCM can represent the gray-scale pair probability statistics information of all pairs of pixels with different step lengths and different directions in the entire image, but in practical applications, resear...
View more


The invention discloses an image texture feature extraction method based on a nested triangular structure, relating to a brand new gray different triangle matrix (GDTM). By extracting the gray difference of the nested triangle vertexes of the image at different positions and angles, counting the frequency of the gray difference, and performing normalization calculation to obtain a triangle matrix. The GDTM can reflect the distribution features of the brightness change, can reflect the comprehensive information of the image in different directions and different interval brightness changes, and is the image brightness change second-order statistic characteristic based on a triangle geometry structure. The texture characteristic quantity calculated on the basis of the invention can visually and effectively represent the image texture condition, compared with the conventional texture statistic feature extraction method such as the grey level cooccurrence matrix, the image texture feature extraction method provided is less in calculated quantity, is resistant to illumination change and RST change, can improve the description effect of large textures, and is an effective image texture feature extraction method.

Application Domain

Image analysis

Technology Topic

Feature extractionBrightness perception +3


  • Image texture feature extraction method based on nested triangular structure
  • Image texture feature extraction method based on nested triangular structure
  • Image texture feature extraction method based on nested triangular structure


  • Experimental program(1)

Example Embodiment

[0060] Examples:
[0061] Such as figure 1 As shown, the flow chart of a new image texture feature extraction method GDTM. In order to evaluate the stability and discrimination ability of the features obtained by the method, the simulation experiment of the embodiment uses the UIUC texture library. The library contains 25 types of textures, each of which contains 40 A 640×480 pixel gray texture picture, Figure 4 This is an image example of this embodiment. In the experiment, 30 of the 7 pictures are selected as training pictures, and the remaining 10 pictures are test pictures. Figure 5 In order to calculate the specific execution process of a picture GDTM matrix, the specific steps of the embodiment are as follows:
[0062] Step 1: Preprocess each original texture image in the UIUC library, and quantize the gray level of the image to G gray level. In this embodiment, G=16.
[0063] Step 2: Scan the preprocessed image pixel by pixel in the order of rows and columns, and obtain the vertices of each layer of triangles in all nested triangle structures with each pixel rotated by 0°, 90°, 180°, and 270° according to the central axis. Correspondingly, a three-dimensional coordinate matrix C=[axis vertex coordinates, left vertex coordinates, and right vertex coordinates]. In this embodiment, C is not layered.
[0064] Step 3: Let x and y be the coordinates of the pixel point P, and the grayscale image function is f(x,y), then f(x,y) is the grayscale value of P, and calculate f(C). Calculate the gray difference between the axis vertex, the left vertex, and the right vertex two by one, and take the maximum and minimum values, then each triangle corresponds to a maximum and minimum gray difference pair, so as to obtain the gray difference pair vector D = [(Maximum Grayscale Difference and Minimum Grayscale Difference)]
[0065] Step 4: The gray level of the image is L, the gray level difference range is 0~L-1, the number of occurrences of gray level difference pairs is counted, and a matrix of L×L size is obtained, and normalized calculation is used to obtain the GDTM gray level difference pair probability Triangular matrix, the size of the GDTM matrix in this embodiment is 16×16. Image 6 It is an example of the gray level difference versus probability triangle matrix GDTM of a picture calculated through the above steps.
[0066] Step 5: Derive the image texture features on the basis of the GDTM matrix obtained in Step 4. In this embodiment, five texture features are calculated with reference to GGCM, which are energy T 1 , Contrast T 2 , Inverse moment T 3 Entropy T 4 And autocorrelation T 5 , Its calculation formula is as follows, where the matrix GDTM obtained in step 4 is denoted as G:
[0071] among them
[0072] Figure 7 Is part of the feature data of the exported 5 texture features; for more intuitive analysis of feature data, Figure 8 Shows 5 texture characteristic curves based on GDTM of 280 pictures (7×40 types) in the specific embodiment;
[0073] Picture 9 For a specific embodiment of the present invention, the classification accuracy of various features on the UIUC texture library is compared; where LTP is a local three-valued mode, CLBP is a complete local binary mode, CLBC is a complete local binary calculation, and the classifier uses a support vector machine SVM. The results show that compared with other methods, the GDTM classification effect is better. Although the accuracy rate still has a large room for improvement, this embodiment is only the simplest and basic execution process of the present invention, and the results obtained are omitted compared with other methods. There are advantages.


no PUM

Description & Claims & Application Information

We can also present the details of the Description, Claims and Application information to help users get a comprehensive understanding of the technical details of the patent, such as background art, summary of invention, brief description of drawings, description of embodiments, and other original content. On the other hand, users can also determine the specific scope of protection of the technology through the list of claims; as well as understand the changes in the life cycle of the technology with the presentation of the patent timeline. Login to view more.

Similar technology patents

Flame retardant phenolic resin foamed sandwiching material and its preparation

InactiveCN1613638ASuppresses fume and hydrogen halide formationgood stability

Exterior wall coating material capable of finishing working through one-time painting

ActiveCN105754431Agood stabilityGood alkali resistance

Vision-based road detection method

InactiveCN102156977Agood stability
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products