Application of the hottest threshold method in ima

Application of threshold method in image segmentation technology

image research so in order to ensure the personal safety of operators, people are often only interested in some parts of the image, which are called targets or prospects (other parts are called backgrounds), and they generally correspond to specific and unique areas in the image. In order to identify and analyze targets, they need to be separated and extracted, on this basis, it is possible to further use the targets. Image segmentation refers to the technology and process of dividing an image into regions with pattern characteristics and extracting interested objects. Here, the characteristics can be the grayscale, color, texture, etc. of the pixel, and the predefined target can correspond to a single region or multiple regions. The existing image segmentation algorithms include threshold segmentation, edge detection and region extraction. This paper focuses on image segmentation technology based on threshold method

1 threshold method image segmentation

1.1 basic principle of threshold method

threshold segmentation method is a region based image segmentation technology. Its basic principle is to divide image pixels into several categories by setting different characteristic thresholds. Common features include: gray or color features directly from the original image; Features obtained from the transformation of original gray or color values. Let the original image be f (x, y), find the eigenvalue t in a certain criterion f (x, y), and divide the image into two parts. The segmented image is:

if b0=0 (black), b1=1 (white), that is what we usually call image binarization

1.2 threshold method classification of image segmentation methods

global threshold method refers to the use of global information to calculate the optimal segmentation threshold for the whole image, which can be single threshold or multi threshold; The local threshold method divides the original whole image into several small sub images, and then applies the global threshold method to each sub image to calculate the optimal segmentation threshold. The global threshold method can be divided into point based threshold method and region based threshold method. The result of threshold segmentation method largely depends on the selection of threshold, so the key of this method is how to select the appropriate threshold. Because the local threshold method still needs to use the global threshold method, this paper mainly studies the point based threshold method and the region based threshold method in the global threshold method. According to the principle of threshold method, the threshold selection technology can be divided into three categories

(1) point based global threshold method

compared with other methods, the point based global threshold algorithm has lower time complexity and is easy to implement, which is suitable for real-time image processing systems

(2) region based global threshold method

for an image, different regions may be related to the documentary plastic Kingdom, such as the target region or background region. The pixels in the same region have strong consistency and correlation in position and gray level at the same time

(3) local threshold method and multi threshold method

local threshold (dynamic threshold) when the image has the following conditions: shadows, uneven illumination, different contrast everywhere, sudden noise, background gray changes, etc., if only a fixed global threshold is used to segment the whole image, the segmentation effect will be affected because it can not take into account the situation of all parts of the image. One solution is to segment each part of the image separately with a set of thresholds related to the position of the image cable (i.e. the function of the threshold coordinates). This coordinate related threshold is also called dynamic threshold, and this method is also called change threshold method, or adaptive threshold method. This kind of algorithm has large time complexity and space complexity, but it has strong anti noise ability, and has a good effect on some images that are not easy to be segmented with global threshold

multi threshold method it is obvious that if the image contains several targets occupying different gray level areas, it needs to use multiple thresholds to separate them. In fact, multi field value segmentation can be regarded as the generalization of single threshold segmentation

2 threshold method

the maximum interclass variance method is simple, stable and effective, and has been widely used. It is a popular threshold selection method. The basic idea is to divide the histogram into two groups at a certain threshold. When the variance of the two groups is the largest, the threshold is obtained. Because variance is a measure of the uniformity of gray distribution, the larger the variance value, the greater the difference between the two parts of the image. When some targets are misclassified into backgrounds or some backgrounds are misclassified into targets, the difference between the two parts will become smaller. Therefore, the segmentation with the largest variance between classes means the smallest misclassification probability

the gray of the image is applicable to the tensile experiment of rubber, plastics, textiles, waterproof materials, wires and cables, ropes, wires, metal rods, metal plates and other materials. The degree range is 0, 1, 2,..., L-1. If the pixel points of gray level I are set, and the test speed is too fast or too slow, the test results will be different. The number of samples is Mi, and the total number of pixel points of the image is, then the occurrence probability Pi of gray level I is defined as

in the Ostu method, the threshold T divides the pixels of the image into two categories (c0= (0, 1,..., t) and c1= (t+1, t+2,..., L-1) (representing the doormark and background respectively)

that is, the threshold T divides the image into two parts: the target and the background, so that the T with the maximum total variance of the two types is the best segmentation threshold. (end)