Abstract:In order to measure the crop geometrical feature real-timely in the complex background, interactive software system and hardware system were designed. Based on the Matlab image process technology, crops geometrical feature calculation models and the idea of human-computer interaction and classification calculation, the software system were designed. The whole measuring process was divided into four steps, which were image acquisition, RGB color feature acquisition, crops region separation from the background and crops geometrical feature calculation. A human-computer interactive software interface was designed, which consisted of menu bar, operation button, image display window and measurement value display window, and it could control each measuring step. RGB color eigenvalues of different crops were obtained by using discrete sample point interactive selection method and region interactive selection method, then the crops region was separated from background efficiently by using screening method based on RGB color eigenvalue difference of each pixel. The software system built classified color database for some kinds of crops according to the environmental condition and crops feature, so pertinence and efficiency for extracting crops region were improved, and real-time measurement was realized. Image was rectified to enhance measurement precision. Corresponding relationships between the actual sizes represented by unit pixel and different object distances were determined previously in order to calculate the actual crops geometrical feature values in real time, so it was not needed to set reference object to calculate the actual size represented by unit pixel when the system was used. The software system built calculation models for several crops geometrical features, including volume, area, and length calculation models of revolving and non-revolving body. The crops volume and projective area regressive model was explored, which provided a new idea for calculating the volume of the crops with non-revolving feature. The objectives of classification design were to expand the measuring types of crops, make the calculation process convenient and enhance the calculation speed of the system. The hardware system consisted of high-performance PC core component, industrial touch screen with a rapid responding speed, industrial camera which had the function of automatic focusing in a fixed focal length and adjustable support. High-capacity lithium battery supplied power for the hardware system. The whole system was 29.8 cm×19.9 cm×6 cm in size, and it was easy to manipulate the system, thus the system could be used as a portable device. For the purpose of installing the device on machine, the adjustable camera support was designed, which was made up of electric dividing plate, stepper motor and S7-200 PLC, and industrial camera shooting direction could be controlled by PLC. The measuring accuracy and real-time feature of system were verified, and the result showed that the measuring error was 4%～8%，and the average processing time was 1.63 s (without the time to construct the color feature database). The system could accomplish the measurement of crops geometrical feature with relatively high speed and calculating accuracy in real time. The system could be used in the field of crops geometrical feature measurement.
