| Transplanting is one of the most important operations during vegetable and flower plant production.In order to meet the growing space and nutrient demand of plug seedlings,it is necessary to transplant them from high density to low density.At present,it is difficult to enlarge the production scale because of the lower efficiency and higher labor intensity to distinguish healthy and non-healthy seedlings by hand.So,it has become a crucial task to identify healthy seedlings before transplanting them to achieve high-quality transplanting.In addition,non-destructive plant growth parameters measurement is an important concern in automatic-seedling transplanting.Furthermore,image processing techniques are fast,timesaving and provide a guarantee for the quality,it also has significance for improving efficiency,reducing labor intensity,assuring the transplanting speed and promoting rapid seedlings production development.Recently,several image-based monitoring approaches have been proposed and potentially developed for several agricultural applications.However,the existing recognition techniques based on depth information are mainly realized seedlings identification by 3D reconstruction of the point cloud,image feature extraction or fusion with RBG information,they are much more complex and very limited for the recognition of healthy seedlings.In view of the shortcomings,this thesis was proposed to develop the healthy seedling recognition method based on depth information with Real Sense sensor for the greenhouse seedling transplanting robot.However,the main research contents are as follows:1.Based on the seedling index,the correlation and grey correlation analysis between the morphological characteristic’s parameters and the seedling index of cucumber seedlings were studied According to the analysis,the threshold values of the leaf area,stem diameter and plant height of cucumber seedlings were determined.It was concluded that when the leaf area,stem diameter and plant height characteristics of cucumber seedlings greater than 257 mm2,1mm,and 27 mm,respectively,thus,they could be transplanted.However,if the leaf area,stem diameter and plant height less than 257 mm2,1mm and 27 mm,respectively,thus,they would not meet the requirements of the transplanting.It is recommended that these threshold values could provide the theoretical basis for the feature extraction of healthy cucumber seedlings.2.In order to achieve the close-shot identification and preprocessing of seedling objects,the Real Sense-based machine vision(MV)system for the close-shot seedling-lump integrated monitoring was developed.The strategy was based on the close-shot depth information.Furthermore,the point cloud clustering and suitable algorithms were applied to obtain the segmentation of 3D seedling models.In addition,the data processing pipeline was developed to assess the different morphological parameter of 4 different seedling varieties.The experiments were carried out with 4 different seedling varieties(pepper,tomato,cucumber,and lettuce)and trained under different light conditions(light and dark).Moreover,analysis results showed that there was not significantly different(p<0.05)found towards light and dark environments due to close-shot near-Infrared(IR)detection.However,the results revealed that the stem diameter relationship between Real Sense and the manual method was found for 2=0.68 cucumber,2=0.54 tomato,2=0.35 pepper,and 2=0.58 lettuce seedlings.Whereas,the seedling height relationship between Real Sense and the manual method was found higher than 2= 0.99,0.99,0.99,and 0.99 for pepper,tomato,cucumber,and lettuce,respectively.Based on the experiment results,it was concluded that the RGB-D integrated monitoring system with the purposed method could be practiced for nursery seedlings most promisingly without high labor requirements in terms of ease of use.The system revealed a good sturdiness and relevance for plant growth monitoring.3.In order to further verify the validity and versatility of the close-shot recognition method,the designed algorithm was applied on the cucumber seedlings to identify healthy seedlings during transplantation mode based on obtained threshold values.The experiment was conducted on the developed full-automatic integrated transplanting machine system and identified the non-grabbed seedlings,inferior seedlings,and healthy seedlings one by one.The experimental results displayed that the recognition accuracy of 105 holes cell trays was 96.59 % and taken time was 0.3 second on the 10 th day.It was observed that the recognition rate of healthy seedlings was decreased with increases of the growth stage of seedlings.However,on the 16 th day,the recognition rate of healthy seedlings was 92.65 %.Therefore,based on the experiment results a comprehensive evaluation method of healthy seedlings based on multi-characteristics was established.In summary,Real Sense SR300 was taken as a depth information acquisition equipment for the advantages of low-cost,high-precision and small integration.Taking advantage of simple structure in close-shot vision,the healthy seedling recognition method of cucumber seedlings during transplantation mode was realized.The whole algorithm is much simpler,which can eliminate the unhealthy and inferior seedlings significantly.It has good feasibility for many kinds of seedlings.Additionally,this research has the perspective for future practical value to real-time vision servo operations for identification of different types of healthy seedling during transplanting on transplanting robots. |
PDF Full Text Request