CAJ | 학술논문

　　边缘检测是图像处理和计算机视觉中的基本问题。为了能有效地捕获目标的主要特征，提出了一种基于Ostu阈值分割和数学形态学的灰度图像边缘检测新算法。利用Ostu算法找到一个最佳的阈值，根据这个阈值把灰度图像二值化，构造四个不同方向的3×3或5×5的结构元素模板，采用数学形态学中的腐蚀算法利用元素模板来腐蚀二值图像。自适应地均衡腐蚀结果，检测出图像边缘。仿真和分析表明同传统边缘检测算法相比新算法运算量小，检测的边缘轮廓清晰，连续性好，可以很好地提取图像中富含的边缘信息，且抗噪声性能较好，假边较少，适用性强。
　　변연검측시도상처리화계산궤시각중적기본문제。위료능유효지포획목표적주요특정，제출료일충기우Ostu역치분할화수학형태학적회도도상변연검측신산법。이용Ostu산법조도일개최가적역치，근거저개역치파회도도상이치화，구조사개불동방향적3×3혹5×5적결구원소모판，채용수학형태학중적부식산법이용원소모판래부식이치도상。자괄응지균형부식결과，검측출도상변연。방진화분석표명동전통변연검측산법상비신산법운산량소，검측적변연륜곽청석，련속성호，가이흔호지제취도상중부함적변연신식，차항조성성능교호，가변교소，괄용성강。
The edge detection is the basic problem of image processing and computer vision. In order to capture objects’main features effectively, this paper puts forward a novel edge detection algorithm using Ostu’s threshold selection method and mathe-matical morphology for gray-level images. The new algorithm uses Ostu’s method to select a optimal threshold, transforms gray image to binary image based on the threshold, then constructs four structuring elements of 3×3 and 5×5 templates of different directions, uses these templates to erode the binary image based on mathematical morphology. It balances the erosion results adaptively and detects the edge. The results and analysis indicate that the new algorithm is simple and detects a clear and continuous edge, extracts much edge information included in the images, meanwhile, the anti noise performance is better than the classical algorithm as well as false edges are less and it is great applicability.