摘要
针对目前大米外观裂纹指标检测方法落后、检测结果缺乏客观性,可重复性差等缺点,提出一种基于双正交B样条小波变换的大米裂纹检测方法。通过图像增强、图像滤波等步骤从原始图像中提取单体米粒图像。并对提取出的单体米粒图像进行小波变换以突出米粒裂纹特征,大大提高了裂纹检测的定位精度。结果表明,该方法可以比较全面地检测出图像中的实际裂纹,检测边缘位置和实际边缘位置十分接近,并且避免了单个边缘产生多个响应的情况,为进一步完善大米品质检测提供了理论和实践基础。
The rice fissure detection method based on dual orthogonal 1~ -3pllne mum -resomt,o~, wav~ transformation has been developed to solve the problems of low efficiency,test results lack of objectivity and poor re- peatability by using traditional manual method. Single rice images can be obtained through the methods of images en- hancement and images filtering from original images. Using wavelet change of extracts of singe rice iamge highlights the features of rice fissure. And it greatly improves the location accuracy of fissure detection. The experiment result shows that this method can quite comprehensively detect the actual fissure in the images, and image edge of detection is close to the actual image edge. Besides, it also avoid the situation that single edge produces many responses, and it provides foundation of theory and practice for further improvement of the quality testing of rice.
出处
《中国粮油学报》
EI
CAS
CSCD
北大核心
2012年第6期101-104,共4页
Journal of the Chinese Cereals and Oils Association
关键词
大米
裂纹检测
多分辨率
B样条
小波分析
rice, fissure detection, multi - resolution, B - Spline, wavelet analysis
