Research On Detection And Classification Index Of Cut Rose Based On Image Recognition

Classification is a key link to ensure the quality of cut rose,however,there are still certain problems to be solved in the classification of cut rose in China.On the one hand,most of the cut roses in China are classified by manual selection currently,which is inefficient and inaccurate,on the other hand,the standards of cut roses in China mostly use descriptive terms to evaluate the grades of each part of cut roses,which lead to some ambiguity in the classification of cut rose.The quality of cut roses was studied based on image recognition technology in this paper,in order to improve the efficiency and accuracy of the classification of cut roses.The main works of this paper are as follows:(1)According to the classification index of cut rose,the flower part and the stem part of cut rose were selected as the main research objects of this paper.Two special image acquisition boxes were built to acquire the side view projection and the vertical view projection images of rose flowers and the side view projection images of the whole cut roses.Selected the appropriate noise elimination algorithm for the images of cut roses,according to the characteristics of different projections of cut roses,selected the appropriate image segmentation algorithms and morphological processing methods to remove the backgrounds and the shadows from the images.(2)Detected the blooming degree,size and fullness of rose flower.Extracted rose flowers from the side view projection images of rose flowers.According to the characteristics of morphological of rose flowers in different blooming degrees,a model of rose blooming degree was constructed,extracted the blooming degree characteristics of rose flowers by scanning the convex hull of the contour of a rose flower from its blooming base point,then optimized the results of scanning,experiments showed that this method can acquire the characteristics of rose flowers blooming degrees accurately.For the rose flowers with lower blooming degrees,the height of rose flower was extracted as the basis of flower size evaluation,and the accuracy of the method was verified.Extracted rose flowers from the vertical view projection images of rose flowers,according to the characteristics of morphological and size of the rose flowers with higher blooming degrees,the fitting circle of the vertical projection of rose flower contour was extracted by least square method to describe the size of rose flower diameter,according to the morphological characteristics of rose flower,the ratio of the diameter of fitting circle and circumscribed circle was extracted to describe the fullness of rose flower,and the accuracy of two methods were verified.(3)Detected the length and the bending degree of the stem of cut rose.Separated the rose flower and the branch from the side view projection image of the cut rose,extracted the length of the branch,acquired the length of stem accurately by using the height of cut rose to subtract the height of rose flower;removed most of the leaves and lateral branches in the branch image by distance transformation and morphological processing,extracted the central line of stem,used the deflection angle between two ends of the central line of stem to describe the bending degree of stem,the accuracy of the method was verified.(4)Collected the morphological characteristic parameters of images of flowers and stems of cut roses in different grades,compared the parameters with the grades of flowers or stems of cut roses,and the classification standards of flowers and stems were quantified by statistics,statistical and quantitative values of flowers and stems classification indices were acquired.