Fast Image Filtering Method Based on Space Filling Curve and Extreme Points

Abstract

The invention relates to a fast image filtering method based on space filling curves and extreme points. The fast image filtering method based on the space filling curves and the extreme points comprises the following steps: a first step is that a source image to be filtered is input; a second step is that a Hilbert curve is generated; a third step is that pixel points in a two-dimensional image domain are subjected to sequencing according to an order indicated by the Hilbert curve, and a one-dimensional pixel point sequence is generated; a fourth step is that a maximum envelope and a minimum envelope of the one-dimensional pixel point sequence are calculated, and the obtained envelopes are subjected to fine processing; a fifth step is that the maximum envelope and the minimum envelope after the fine processing are used as input values, and an image base and image details after a smoothing filtering process are obtained through calculation of a mean value; and a sixth step is that one-dimensional filtering results of the image base and the image details are subjected to counter-Hilbert sequencing, an original two-dimensional image is divided into a smooth part and a detail part, and therefore filtering is finished. The fast image filtering method based on the space filling curves and the extreme points can extremely reduce interpolation errors and lower the complexity of the filtering, and can achieve a quasi-real-time effect.

Description

technical field [0001] The invention relates to an image filtering method, in particular to a fast image filtering method based on space filling curves and extreme points. Background technique [0002] With the continuous development of image filtering technology, all kinds of new filters provide very useful help for various aspects of image processing, such as digital entertainment, detection and recognition, editing and synthesis. Therefore, continuous improvement of filter performance and expansion of filter application domain have been extensively studied by scholars at home and abroad. [0003] In 2009, Subr, K. published an article entitled Edge-preserving multiscale image decomposition based on local extrema on the top international journal ACMTrans.Graph. A filter based on extreme points was introduced. Different from the idea that traditional filters are usually based on pixel intensity filtering, this filter uses extreme points to define the characteristics of the...

AI Technical Summary

Problems solved by technology

However, since the high-quality fitting of two-dimensional maxima-minimum envelopes is usually time-consuming and has large calculation errors, the practical application of this method is greatly limited.

[0004] In the face of various practical applications, traditional image filters either cannot adapt to a wide range of filtering requirements in terms of effect, or are not ideal in terms of time performance, and cannot meet users' real-time or quasi-real-time filtering needs

Images