The Research Of Several Rice Phenotype Parameters Extraction System

Both the traditional and genetic breeding require phenotypes screening after growing and cultivating. So all the phenotypes should be measured and analyzed systematically so as to screen the rice varieties with high efficiency. The domestic measurements are manually or by single parameter machine. It is time-consuming and expensive not to mention its inaccuracy caused by individual differences. The task should be done in a relatively short period so as to make the data comparable and that challenges the handwork as well. The single parameter machines, on the other hand, have limited functions, and can not perform unified data management. It, to some extent, restricts the development of high throughput analysis of the breeding work. Although some automatic phenotype measurements are being studied abroad lately, they still at an initial stage.The Design of Parameter Automatic Extraction in Rice Growth Duration would realize the high throughput automatic acquisition of phenotype data including plant height, tiller number, filled grain number. The unique trackable label of the sample enables the scientists to follow the special treated rice. A database will be built up after the measurement which facilitates the management and analysis task.This Article covers the following aspects: 1) Comprehensive literature research on the development and important progress in the field of agriculture automatic acquisition; 2) Possible alternatives and the final decision; 3) Analyzing and classification of the solution to setup the framework; 4) Realization of the individual measurement modules and hardware system; 5) Realization of the whole package of solution; 6) Testing and analyzing the measurement of the parameter automatic extraction platform.After 120 hours testing, 21500 rice samples were treated by the System. The average processing time is 20 seconds per plant better than the objective of 30 seconds. Results show that there are good relationships between the plant height instrument result and manual result, with correlation coefficient being 0.998. The relative deviation is 0.38%, and the coefficient of variation is 0.72%. Correlation coefficient between the tiller number instrument result and manual result is 0.977. The relative deviation is 1 tiller number better than the requirement of 2. And there are good relationships between the filled grain number result and manual result, with correlation coefficient being 0.993.. The relative deviations and coefficients of variation is 0.51% and 1.78% respectively. The new idea and innovative point are the realizing of intellectual high throughput measurement of rice tiller number by CT and image processing techniques and the realizing of auto-high throughput measurement of filled grain number by linear array camera and X-ray transmission tomography.This article is an important composition of the automatic acquizition platform. Its development would greatly benefit the rice breeding work, bring about a series of proprietary intellectual property rights and make tremendous economical and social profits.