Enhanced interval type 2 FCM method and image processing system

Abstract

The invention belongs to the technical field of FCM algorithms, discloses an enhanced interval type 2 FCM method and an image processing system. The method comprises calculating a reference clusteringcenter, and arranging clustering center vectors of corresponding features according to an ordering rule of the reference clustering center to obtain an initial clustering center; initializing the clustering center with the weighted average of sample features, which has good adaptability to different types of samples; introducing the ordering of a KM algorithm in the initialization process, and directly substituting an initial central value to search a switching point in a type-reduction process. The method can improve the ability of the algorithm to deal with the uncertainty by the introducing the interval type 2 fuzzy theory, greatly reduces the computational complexity of the interval type 2 FCM algorithm by optimizing the initial clustering center and the type-reduction operation, andimproves the convergence speed of the algorithm. The effectiveness of the improved algorithm is verified by experimental comparison of random and actual data.

Description

technical field [0001] The invention belongs to the technical field of FCM algorithms, and in particular relates to an enhanced interval type II FCM method and an image processing system. Background technique [0002] At present, fuzzy clustering has been widely used in image processing, pattern recognition, computer vision and other fields because of its unsupervised characteristics and easy-to-understand logic language. Among them, the theory of fuzzy C-means clustering (FCM) algorithm based on objective function is the most Perfect and most widely used. The traditional FCM algorithm achieves clustering by iteratively updating the distance and membership degree from the sample to the cluster center, and uses a type-1 fuzzy set with a value in the interval [0, 1] to express the membership relationship between the sample and each cluster center. In practical applications, the clustering process contains various uncertain information, such as uncertainty factors in the proce...

AI Technical Summary

Problems solved by technology

At the same time, consider that IT2FCM randomly selects cluster centers, which increases the number of iterations and affects the final clustering results

[0003] To sum up, the problem existing in the existing technology is: when the traditional FCM algorithm processes data samples or image samples, it cannot accurately obtain the cluster center of the sample, and its processing method is to start iteration with a random cluster center value. The iterative method is very easy to converge to a local optimum, resulting in a decrease in clustering accuracy

On the other hand, the Type II FCM algorithm inevitably increases the complexity of the algorithm while improving the ability to process samples, and the operation speed is difficult to meet the requirements when processing large data samples, and the operation of this type of algorithm involves fuzzification, fuzzy There are four links of reasoning, reduction, and defuzzification. How to simplify the computational redundancy in these four links is also a difficult problem in the practical application of the algorithm.

Images