Hierarchical super pixel segmentation model based on histogram one-dimensional differential distance

Abstract

The invention relates to a hierarchical super pixel segmentation model based on histogram one-dimensional differential distance, and presents an algorithm for calculating similarity based on histogramone-dimensional differential distance. The algorithm is simple and accurate. A pyramid-based super pixel segmentation model and a corresponding hierarchical merging method are established, and different methods are adopted according to the characteristics of different layers, so that the accuracy is not reduced while the operation speed is increased. Experiments show that, compared with the classic algorithm, the algorithm of the invention achieves high accuracy, efficiency and compactness, and has better anti-noise performance.

Description

technical field [0001] The invention belongs to the field of color image segmentation, and specifically designs a histogram-based Figure 1 A Hierarchical Superpixel Segmentation Model with Dimensional Differential Distances. Background technique [0002] With the advancement of technology, high-resolution and large-size images have gradually become mainstream. In the fields of saliency detection and object tracking, the urgent requirement for real-time performance makes the concept of superpixel emerge as the times require. The concept of superpixel was proposed by Ren et al. in 2003, which refers to an image block composed of adjacent pixels with similar texture, color, brightness and other characteristics. [0003] Superpixel segmentation can be classified into two categories based on gradient descent and graph theory. In order to balance efficiency and accuracy, RadhakrishnaAchanta et al. proposed the SLIC algorithm in 2010. The basic idea is local K-means clustering....

AI Technical Summary

Problems solved by technology

However, the compactness is not high, and the segmentation of details is not ideal. Although the anti-noise performance is greatly improved compared with the SLIC algorithm, it is still not ideal.

[0004] In order to solve the problem of poor anti-noise performance of the above algorithm, this paper establishes a pyramid-based hierarchical superpixel segmentation model, and at the same time establishes a new histogram similarity measure based on structural similarity to improve the boundary fit. method, and introduce a compactness constraint term to control the compactness in the segmentation process

Images