The Design And Implementation Of Extraction Of Common Morphological Leaf Prototype System

Plants are one of Earth’s most common life，which play an important role inecological cycle, protecting atmospheric composition and maintaining other naturalsystems. The leaf are elementary organs of plants.so it is a necessarily step to measuringleaf for the research and identification of leaf.On the base of the elementary characteristics of plant leaf and image processing, thispaper introduces the geometric characteristics of leaf and the veins of leaf firstly. Thegeometric characteristics includes in the outline of features, the scale of length and width,angle of base characteristics and angle of apex characteristics and so on. Thecharacteristics of veins mainly means primary veins. Secondly, the algorithms to measurethese features are proposed including in some algorithms of preprocessing, somealgorithms of geometrical features extraction, some algorithms of vein featuresextraction.And then two modules is designed which include in the measurement of geometricfeatures and the measurement of vein features. In the modules of the measurement ofgeometric features we successively use the average threshold of binarization method toextracting binary image,the opening in mathematical morphology to noise reduction, thecanny algorithm to extracting single-pixel-width outline and then enlarging,aspect–ratio-algorithm to computing aspect ratio， tangent algorithm to computingcharacteristic angle to measuring the geometric features. On the other hand, wesuccessively use k-means algorithm to extracting the profile of vein, the positivedirection of the scanning to extracting primary to measuring the vein features. Furtherly,the optimal parameters is proposed.Finally, Based on the classified theory of decision tree，we test two module with208pictures of leaf, getting the characteristic parameters,and classify the testingpictures. We conclude both module can compute the characteristic parameters and thesefeature could be used to classification.