基于灰度差分统计的火焰图像纹理特征提取  

Texture Feature Extraction of Flame Image Based on Gray-Scale Difference Statistics

作　　者：刘辉[1] 张云生[2] 张印辉[3] 何自芬[3] 

机构地区：[1]昆明理工大学冶金与能源工程学院,云南昆明650093 [2]昆明理工大学信息工程与自动化学院,云南昆明650051 [3]昆明理工大学机电工程学院,云南昆明650093

出　　处：《控制工程》2013年第2期213-218,共6页Control Engineering of China

基　　金：国家自然科学基金(No.60962007)~~

摘　　要：转炉吹炼过程中,炉口火焰的纹理粗糙度与吹炼所处的阶段有密切的联系。在不同的吹炼阶段,火焰纹理呈现较明显的变化,因此该特征可以作为转炉终点识别系统的重要参数之一,目的是从火焰图像中提取出描述纹理复杂度特征的描述子,进而区分出不同的吹炼阶段。提出基于灰度差分统计的火焰纹粗糙度提取方法,首先对原始火焰图像做去噪、分割、灰度转换的预处理,然后计算灰度差分直方图,在此基础上对比各种特征描述子表述粗糙度的效果。结果表明,和灰度共生矩阵方法的能量特征值相比,灰度差分统计方法中的角度方向二阶矩描述子对不同纹理所处的吹炼时期具有较高的识别率,且具有较快的识别速度,该特征描述子能作为重要的特征参数之一用于实际的转炉吹炼终点判断识别系统中。In the process of converter blowing, the texture roughness and the stage of blowing have a close relationship. The roughness of flame texture shows a significant change in different stages, this texture feature is the one of the important parameters to the converter end-point forecast identification system. The purpose of this paper is to get the complexity descriptor from the flame images, which can distinguish different blowing stages. The method of texture roughness of flame image based on gray-scale difference statistics is pro posed. Firstly, the pretreatment of de-noising, segmentation and grayscale conversion were done to the flame images, and then the dif ferential histogram was calculated, at last the effect of various features roughness descriptor was compared. The result shows that the di rection angle second-order moment descriptor of differential statistical method has a better recognition rate and a faster recognition speed than energy descriptor of gray level co-occurrence matrix. This descriptor can be an important feature parameter used in the real convert er end-point forecast recognition system.

关 键 词：火焰图像 灰度差分统计 纹理特征 灰度共生矩阵 角度方向二阶矩 

分 类 号：TP27[自动化与计算机技术—检测技术与自动化装置]