Self-adaptive threshold extraction method for large-scale grayscale image

Abstract

The embodiment of the invention discloses a self-adaptive threshold extraction method of a large-scale grayscale image, and the method comprises the steps: reading in a series of large-scale grayscaleimages in a matrix manner, and sequentially carrying out the Wiener filtering and Gaussian filtering on images meeting the requirements, and obtaining an image sample; then counting gray value information of the image sample to obtain a gray scale histogram distribution diagram of the image sample, and reading gray scale maps of a series of images to perform histogram accumulation; solving the gradient according to the accumulated histogram distribution to obtain different gradient distribution data of the image sample; according to the gray scale histogram distribution diagram and the gradient distribution data result of the image sample, solving the gray scale distribution characteristic value solving, and calculating a segmentation threshold value through an empirical formula; and finally, calculating an adjustable range of the segmentation threshold, and providing range limitation for manual adjustment. The method provided by the invention has high adaptive capacity, the whole process is automatically carried out, and the identification and extraction efficiency and the identification accuracy are greatly improved.

Description

technical field [0001] The embodiments of the present invention relate to the technical field of digital image processing, and in particular to a method for extracting adaptive thresholds of large-scale grayscale images with significant four-component features. Background technique [0002] Image threshold segmentation technology refers to the technology of distinguishing different components in an image according to different thresholds. There are two main types of traditional threshold segmentation. One is the manual adjustment method. Through visual recognition, different segmentation thresholds are tested on the image. The optimal segmentation threshold is obtained by adjusting the property; the other is the automatic threshold segmentation method represented by the Otsu method, which is based on the idea of ​​​​the maximum inter-class difference to perform two-component image segmentation. [0003] The manual adjustment method can sometimes obtain a relatively credible ...

AI Technical Summary

Problems solved by technology

[0003] The manual adjustment method can sometimes obtain a relatively credible threshold, but it needs to repeat the process of segmentation test. There are too many subjective judgments in this process, which brings inaccuracy; on the other hand, the segmentation test often takes a long time. Time, not economical and operable; at the same time, the manual adjustment method is operable for small-scale single images, and it is impossible to perform accurate adjustments for large-scale processing objects with a series of pictures

[0004] The Otsu method based on the maximum inter-class difference judgment has relatively good stability and operability, and is an important method for two-component recognition, but it is difficult to use it for large-scale, four-component (abstractly refers to the In the analysis of features with four types of salient components that can be classified by naked eyes), the main reason is that as the matrix increases, the amount of calculation brought is non-linear growth, and the two-component method is used to calculate the threshold of four-component features , will bring repeated iterative process, further increase the amount of calculation, and bring certain inaccuracy

Threshold segmentation is only the first step in image recognition and extraction. The calculation amount brought by the current method will even exceed the recognition and extraction itself.

Images