基于改进Freeman链码的柑橘簇生区域数量判别方法

Quantity Qualification of Overlapped Region for Citrus Image Based on Modified Freeman Chain Code Algorithm

水果簇生区域的数量判别是图像处理技术的难点。提出了一种改进的Freeman链码,即在Freeman8邻域链码的基础上插入了3个新的链码元素"S"、"8"和"9"。识别过程中,首先获得水果簇生区域二值图像的改进Freeman 8邻域链码,然后计数链码中"8"出现的次数,再根据研究对象的轮廓特性,计算出簇生区域的数量。利用该方法对柑橘树数字图像中簇生区域的柑橘数量进行了判别,实验结果表明,该方法对2～5簇生判别效果良好,效率高,其中2、3簇生识别正确率为100%。

A modified Freeman chain code algorithm to determine the number of overlapped regions in citrus binary images was proposed. Three components, including S, 8, and 9, were inserted in the original Freeman code. During the identification process, the improved chain code was first obtained. Then the appeared times of code ＂8＂ was computed. The number of fruits in the overlapped regions can be obtained according to the outline of the fruits. The improved chain code algorithm was applied to evaluate the citrus images. The experimental results showed that the proposed algorithm was effective in identifying the overlapped fruits when the number of overlapped regions was between 2 and 3 where it achieved accuracy of 100%.