Image Texture Main Direction Recognition Method Oriented to Imaging Measurement

Abstract

The invention discloses an image texture main direction recognition method oriented to imaging measurement. Firstly, a texture image is selected, then edge detection is carried out on the texture image through an edge detection operator, texture main direction recognition is carried out on a binary gray image subjected to edge detection, local orthogonal coordinate spaces are projected to the same parameter space through a moving window neighborhood strategy, points on the same line of an orthogonal coordinate system are projected to parameter coordinates and overlapped, all feature points in sub-neighborhoods are voted, values corresponding to extreme points of vote numbers are found in the parameter space, and a weighing average value under a parameter coordinate system is calculated after voting. The image texture main direction recognition method can effectively improve precision of image measurement in actual application, and is easy to operate, little in calculation amount, small in storage load, high in precision and high in robustness.

Description

technical field [0001] The invention relates to an image texture feature extraction technology, in particular to an image texture main direction recognition method oriented to imaging measurement. Background technique [0002] Texture analysis plays an important role in the fields of computer vision and image processing. Especially in imaging-based measurement technology, texture orientation is a very important feature in texture analysis, so obtaining texture orientation information simply and efficiently has become a new research hotspot in the field of imaging measurement. [0003] At present, the general texture direction feature extraction methods are Hough transform, Gabor wavelet, and Radon transform, but these methods have the following technical difficulties: (1) The calculation space and time are large, and the storage burden is large, such as Hough transform, which requires To access each pixel of the picture, the time complexity is approximately equal to the tot...

AI Technical Summary

Problems solved by technology

[0003] At present, the general texture direction feature extraction methods are Hough transform, Gabor wavelet, and Radon transform, but these methods have the following technical difficulties: (1) The calculation space and time are large, and the storage burden is large, such as Hough transform, which requires To access each pixel of the picture, the time complexity is approximately equal to the total number of pixels in the picture, and each pixel must vote for all the lines containing it, forming a one-to-many mapping mode, resulting in blind calculation (2) Low precision, such as Gabor wavelet, when extracting fabric texture direction, the detection accuracy can only reach 1°; (3) Poor robustness, because the detected texture image may contain imaging dark noise , false straight lines appear in the image to be detected, resulting in false detection during detection

Therefore, the above problems restrict the application in engineering

Images