Segmentation method for live pig images in complex environment

Abstract

The invention discloses a segmentation method for live pig images in complex environment. The method comprises the following steps: obtaining the live pig images in real time; establishing a Cartesian coordinate system; conducting gray scale conversion to live pig images to be segmented for gray scale pictures of a live pig followed by a process for the gradient pictures of the live pig; segmenting and extracting for an initially segmented image; performing calculations on the initially segmented image; removing the connected domain for a second segmented image; calculating to obtain an object, the size of the object, the light intensity and the distance; constructing a biased swelling element; calculating the shadow compensation coefficient; processing the second segmented image for a fourth segmented image; and processing in accordance to the fourth segmented image and in combination with the second segmented image for a fifth segmented image. The method of the invention can effectively tell a pig body from a live pig image in a pig house with complicated environment. It can also process consecutively in batches and can be self-adjusted, which makes the method suitable for processing images monitoring live pigs for a long time and helps realize the automatic monitoring to live pigs.

Description

technical field [0001] The invention relates to a pig image segmentation method, in particular to a pig image segmentation method in a complex environment. Background technique [0002] Pig farming has gradually changed from the traditional extensive farming method to a more environmentally friendly and intensive one. In order to make pig farming more efficient, it is necessary to monitor the pigs and detect their physiological status. Traditionally, the monitoring of the status of pigs is carried out manually. Due to the stuffy environment of the pig house and the bad smell, it has had a long-term adverse effect on the health of the staff. In addition, the staff cannot always monitor the pigs. The situation is slow to respond. It is a good choice to replace the manual method by using the machine vision system to obtain the image of the live pig and to detect the state of the live pig through the image processing technology. [0003] In recent years, due to technological ...

AI Technical Summary

Problems solved by technology

However, the environments used in most studies are relatively simple, or have been artificially modified to a certain extent, so as to facilitate image segmentation and subsequent processing, and the data used are often short-term, so that the changes in all aspects of the image are not obvious. The method is also simpler

At present, the environment of many pig farms is complex, and there is no effective image segmentation method to segment the pig body.

Images