扁平颗粒体轮廓提取算法研究与应用

针对扁平颗粒体的质量检测，以西瓜子为例，选取HSI颜色空间，应用阈值构造法构造出符合西瓜子类目标物体特征的一种阈值，将此阈值与Sobel算子相结合，并对提取的基本轮廓优化处理，最终得到高质量的瓜子轮廓，为后续分检过程提供了有力依据。实验结果证明，文中采用的方法简单有效，提取出的轮廓非常清晰，已应用于实时西瓜子自动分检系统，且具有一定的通用性，适合处理类似扁平颗粒体如南瓜子、葵花子等。

Research and Application of Contour Extraction Algorithm for Flat Objects

Based on HSI model, a kind of threshold, which complies with characteristics of melon seeds, has been designed with threshold - making method for quality detection. It has been used to optimize Sobel operator and remove effect caused by non- uniform of image grey distribution. Using parallel contour extraction algorithm for basic contour has made high quality of melon .seeds contour. Experimental results show that not only this method is simple and efficient, especially for real - time environment, but also can be used for other kinds of seeds, such as pumpkin seeds, sunflower seeds in quality detection.