An Image Segmentation Method for Caged Layers Based on Improved Active Contour Model

Abstract

The invention discloses an image segmentation method for caged laying hens based on an improved active contour model. Extract the S-component image of the image of caged layer hens, use the k-means clustering method to divide the S-component image pixels in the square local window into two categories, traverse the S-component image to obtain two class center functions, and construct a The energy function of the function, and the average kernel function and the level set function are introduced into the energy function to form the total energy functional, and the standard gradient descent method is used to minimize the total energy functional to obtain the evolution equation of the boundary line, and the morphological opening operation and Gaussian filtering are added. Finally, the finite difference method is used to iterate the evolution equation of the boundary line until it converges, and the final evolution boundary line is the segmentation result of the image of caged layer hens. The invention can quickly and accurately segment the image of caged laying hens, and eliminate the influence of cage occlusion.

Description

technical field [0001] The invention belongs to the technical field of image processing, and in particular relates to an image segmentation method for caged laying hens based on an improved active contour model. Background technique [0002] Chicken behavior is an important basis to reflect its health, and it is also one of the evaluation indicators to measure breeding welfare (Lao Fengdan et al., 2017). The rapid and accurate segmentation of chicken images is a key step in the application of machine vision systems to quickly identify sick and dead chickens in the actual breeding environment (Bimina et al., 2016). At present, the breeding mode of laying hens in my country is mainly based on the cage mode. Compared with other breeding modes, the laying hens in the cage mode suffer from greater physiological and psychological pressure, and their welfare and health status are more serious. However, due to the occlusion of the cage and the large change in the body shape of the ...

AI Technical Summary

Problems solved by technology

In 2001, Chan and Vese proposed the famous piecewise constant active contour model—CV model, but it is difficult to deal with images with uneven gray levels

In 2008, Li et al. proposed a region shrinkable fitting active contour model—the RSF model, which can effectively segment images with uneven gray levels, but it is sensitive to the setting of the initial contour and is time-consuming.

In terms of agricultural engineering applications, Ma Li et al. (2015) proposed a sow infrared image segmentation method combining CV model and RSF model, but its segmentation efficiency is low, and the anti-occlusion ability and robustness of motion tracking are poor

Xiao Linfang et al. (2018) used the improved C-V model based on morphology to segment the image of caged laying hens, but it needs to use rough segmentation as the initial contour, and the result of rough segmentation will affect the accuracy and efficiency of image segmentation

In summary, the existing segmentation methods based on the active contour model have problems such as sensitivity to initial contour settings, poor motion tracking ability, weak anti-occlusion ability, and time-consuming segmentation.

Images