An Image Segmentation Method Based on a New Particle Filter Algorithm

Abstract

The invention provides an image segmentation method based on a novel particle filtering algorithm, which comprises the steps of building a corresponding dynamic space model in allusion to characteristics of an image to be segmented; acquiring training images, performing preprocessing on the training images, and extracting a region of interest; taking pixels as sample units, selecting network input characteristics, and building a training sample; building a deep neural network model, completing deep network training, and inputting a test sample to acquire an initial segmentation result; generating a particle swarm by using the initial segmentation result, enabling particles to move to a high likelihood region by adopting a particle swarm optimization algorithm, and using an acquired result to act as suggested distribution of particle filtering; and estimating a state variable by adopting a new particle filtering algorithm of the suggested distribution, and acquiring a final image segmentation result. According to the invention, deep learning and the particle swarm optimization algorithm are adopted to generate the suggested distribution, so that a problem of particle degeneration is effectively solved, a good image segmentation effect can be acquired, and the image segmentation method has high applicability and robustness.

Description

technical field [0001] The invention relates to the technical field of image processing and intelligent algorithms, in particular to an image segmentation method based on a novel particle filter algorithm. Background technique [0002] Particle filter implements recursive Bayesian filter through Monte Carlo simulation method, which is suitable for any nonlinear system that can be described by state space model. Because of its simplicity and easy implementation, it has been widely used in target tracking, signal processing, automatic control, and image segmentation. However, the traditional particle filter uses the transition probability density function to generate the proposal distribution, and does not consider the information provided by the latest observation data. There is a certain deviation between the sample drawn from it and the sample generated by the real posterior distribution, resulting in the degradation of the particle. Contents of the invention [0003] Th...

AI Technical Summary

Problems solved by technology

[0003] The purpose of the present invention is to overcome the deficiencies of the prior art, and propose an image segmentation method based on a novel particle filter algorithm, which converts the image segmentation problem into a state estimation problem, and the algorithm adopts deep learning and particle swarm optimization algorithm to generate suggestion distribution, which can effectively solve the problem of particle degradation in filtering, obtain more accurate state estimation, and realize image segmentation

Images