The Study About Detecting Accuracy Of Feeding Device For Drills Based On Machine Vision

Aimed at many current methods in detecting the proformance of feeding device for drills couldn't satisfy the detectning accuracy request nicely, a kind of method which was based on machine sense of vision was put forward for detecting proformance of feeding device for drills. A certain detecting accuracy's analysises were also studied.The method was proved to be a kind of method which was time-saving, labor-saving and needn't touching, having important meaning in the proformance detecting of feeding device for drills.The function of feeding device for drills was determined primarily by the frequency distribute of drill count, the ability of drillng and the even trait in drilling. So seeds count and their distribute were made as main research object respectively. And the relate of sow-drill'number and their distribution provided a theory base for research on proformance detecting of feeding device for drills.An detecting system for feeding device for drills' accuracy was established according the northern production condition in our country, wheat, soybean and corn were collected as the trial samples, experiment project was established according to combination and put into practice, each combination experiment was repeated three times, the detecting accuracy was increased effectively.A machine vision sense detecting method was made by using of the MATLAB. It mainly included: ①The sample was incepted by application of media software, and media file was transformed from MPEG to AVI.② True color sample picture was transformated to ash picture.③Choosing fit picture as background picture in whole picture preface row, adopting picture matrix difference to do away with the object such as machine hull and ruler, then attaining picture with only seeds.④Selecting fit value to deal with the picture, adopting two worth wave demention to do away with the noise ,both attaining the purpose of getting rid of noise and keeping picture edge information, increasing the picture handling speed; ⑤ Adopting procedure language to statistic seed counts in the picture, manifesting end detection result.Through handicraft to statistic the sample seed counts' frequency distribution, compairing with the proceeding by calculator procedure, the results were as follows: The absolute error margin of wheat seed count's distribute was at 0.016~0.116 grains / s ; opposite error margin was between 33% ~ 600%;The absolute error margin of soybean was at 0.004~0.68 grains / s; opposite error margin was between 6.5%~130%;The absolute error margin of corn was at 0.006~0.57 grains/ s; opposite error margin was between 17% ~ 240%. When wheat sample was at high work length or roll speed, the output error margin was bigger, the accuracy of vision sense detecting was lower, The question might lied in noise, the error margin, vibration, environmental interference, the machine control system, picture handles and the data proceeding etc.According to the relation between the count and distribution of row drill .comparing the procedure and handicraft handles in all drill's worth, variation coefficient .detecting the obvious difference with t-statistic ,result is :In wheat sample , The difference of coefficient of variation wasn't obvious , the level of obvious difference were distinguishly 9.9451e-005 and 6.0993 e-004.So they could satisfy the detecting accurance request in wheat row growing. In soybean and corn sample, the average value, standard...