基于机器视觉的瓶装白酒杂质检测

Bottled liquor impurities detecting based on Machine-Vision

摘要:

介绍了一种基于机器视觉的瓶装白酒杂质检测方法.为了从复杂的视觉图像中提取出微小目标,设计了一种倒置翻转式的检测方式,通过高清数字摄像机获取瓶装酒液的视频序列,采用一种改进的二次差分方法,获取瓶装酒液内可能存在的运动目标.根据气泡和可见异物的形态特征进行分类,判断出酒液中是否含有杂质.实验验证了所提方法的有效性.

Abstract:

A method for detecting bottled liquor impurities based on machine vision was introduced. In order to extract small detection targets from a complex image, an inverted flipping detection method was designed. A high-resolution digital video camera was used to acquire the video sequence of the bottled liquor. After that, an improved seconddifference method was utilized to detect possible tiny moving targets in the liquid. Morphological characteristics of the bubbles and visible impurities were used to estimate whether the liquor contains impurities. The experiments demonstrate the effectiveness of the proposed method.