基于小波变换的多尺度边缘检测用于稻谷内部损伤的检测

Detecting Rice Internal Damage by Multi-Scale Edge Detection Based on Wavelet Transform

针对稻谷在成熟、收获、分级包装、装卸运输、烘干、贮藏、加工和销售等各个环节容易产生损伤,而稻谷的损伤程度又难以精确评价的问题,开发了一套稻谷损伤检测装置,通过体式显微成像系统等获取损伤稻谷的高清晰显微图像,基于信号与噪声在不同尺度下小波系数模的变化规律,利用小波变换系数模局部极大值来提取稻谷内部损伤特征信息。结果表明:采用基于小波变换的多尺度边缘检测方法能较准确地提取稻谷内部损伤特征,并有效地降低噪声,较传统边缘检测算子有更好的效果。

Damage is easy to occur for rice kernels during the process of maturation, harvesting, classification, packaging, transportation, processing, preservation and selling, etc., and the precise detection of the internal damage is a difficult subject. In this work, the method for detecting rice internal damage using wavelet transform based on multi - scale edge detection was proposed and the according set was developed. The following is the proposed detection work： First, legible image with high quality are obtained by microscope system; based on different properties of the signal and noise in the wavelet coefficient modulus under different scale of wavelet transformation, detecting the local maximum of the wavelet transformation coefficient modulus, information of the image boundaries feature is provided, and using this property the rice internal damage can be detected. Experimental results show that the multi - scale edge detection for image of rice kernel internal damage is more efficient than the conventional ones.