CAJ | 학술논문

稻米图像边界是阶梯型边界,在稻米图像的边缘检测中,利用模角分离的小波变换,结合尺度独立的算法提取了阶梯型边界,但得到的边缘图像存在边界不连续和边缘响应次数不唯一的问题.而小波变换的模极大值法检测出了所有类型的边界,其中包括不需要的边界.本文提出一种边缘检测方法,将尺度独立法与模极大值法的优点结合,在稻米图像的边缘检测中取得了理想的结果.
도미도상변계시계제형변계,재도미도상적변연검측중,이용모각분리적소파변환,결합척도독립적산법제취료계제형변계,단득도적변연도상존재변계불련속화변연향응차수불유일적문제.이소파변환적모겁대치법검측출료소유류형적변계,기중포괄불수요적변계.본문제출일충변연검측방법,장척도독입법여모겁대치법적우점결합,재도미도상적변연검측중취득료이상적결과.
Rice image borders are step-structure. In the edge detection of rice image, we extract the step-structure boundary using modular-angle-separated wavelets transform combined with scale-independent algorithm. But some borders of the edge image is not continuous, and the number of edge response is not unique. The modulus maximum value of wavelet transform detect all types of boundary including unneeded boundary.The article presents an edge detection method which combines the advantage of scale independent method and modulus maximum value method, and get the desired results in the edge detection of rice image.