摘要
为进一步改善火焰图像边缘检测性能,提出了一种基于复Contourlet模极大值和改进数学形态学的边缘检测方法.通过复Contourlet变换将火焰图像分解成低频和高频子带,并对边缘细节较少的低频子带分量采用改进的数学形态学方法进行处理,检测出较弱的边缘并保证定位准确,而对高频子带分量采用复Contourlet模极大值法检测边缘,可有效减弱噪声,最后经融合得到火焰图像的完整边缘.结果表明:与Canny方法、小波模极大值方法、Contourlet模极大值方法和Contourlet形态学方法相比,所提方法在主观视觉效果、峰值信噪比等客观定量评价指标2个方面有明显的优势,具有更好的边缘检测效果,边缘轮廓更清晰完整,且具有很强的抗噪性能.
To furlher improve the performance of edge detection for flame images, a method of edge detec- tion based on complex Contourlet transform and improved mathematical morphology was proposed. First- ly, the flame image was decomposed into low-frequency and high frequency components through complex Contourlet transform. Then the low frequency sub-band coefficients with less edges and minutiae were processed using improved mathematical morphology so as to detect the weak edges and ensure accurate lo- cation, while the high-frequency sub band coefficients with more edges and minutiae detected using the method of complex Contourlet modulus maxima, with which noise can be suppressed effectively. Finally, the complete edge of flame image was obtained by fusion. Results show that the proposed method has sig- nificant advantages in terms of subjective visual effect and objective quantitative assessment indexes such as peak signal to noise ratio （PSNR） and so on, clearer and more complete edge contour and very strong anti- noise performance, in comparison with Cannny method, wavelet modulus maximum approach, Contourlet modulus maxima and Contourlet morphology method.
出处
《动力工程学报》
CAS
CSCD
北大核心
2014年第1期32-38,共7页
Journal of Chinese Society of Power Engineering
基金
国家自然科学基金资助项目(60872065)
华中科技大学煤燃烧国家重点实验室开放基金资助项目(FSKLCC1001)
江苏高校优势学科建设工程资助项目
