Font Size: a A A

The Algorithm Study Of Image Processing For Mature Rice Based On Video Mining

Posted on:2012-11-24Degree:MasterType:Thesis
Country:ChinaCandidate:K WangFull Text:PDF
GTID:2233330395981791Subject:Agricultural mechanization project
Abstract/Summary:PDF Full Text Request
At present the grain loss rate of Rice Combine generally ranges from2%to4%and even in some large combined harvesters the grain loss rate is above10%.If we can reduce the loss rate of1%.At present the rice area is about4.3million mu in China, the per mu yield is500kilograms, we will increase the rice planting area of0.043million mu. If the price of rice is1.6Yuan per kg, we can the loss of34.4million Yuan. The main reason of the grain loss is the changing rice density during maturation. Rice Combine can not be fit in these changes in real-time.Based on the traditional image processing algorithm, it takes1S to calculate in handling on a frame and the algorithm can not fulfill the real time requirement of the rice combined harvesters Rice density during maturation is connected with rice varieties, there are many varieties of rice and new varieties are constantly emerging. First, It is worth to make a deeply study for how to use an algorithm to adapt to such changes Second, even if the same variety, image processing algorithm depends on the illumination and crop growth situation and other environmental factors Studying a kind of algorithm which can fulfill the real time requirement of the rice combined harvesters and the algorithm will have little to do with rice varieties, illumination and other environmental factors, and it is the problems needed to solve especially.Firstly, it takes the template algorithm to solve the idea of algorithm that is unrelated to rice varieties, illumination and other environmental factors. The main process is as follows:(1) In the experimental plot selects4-6block area at random which have the same size as the area of rice with experimental calibration, the previous method is used to collect and calibrate the area.(2) We can use the mean and the mean of frame differencing as template argument on the basis of the collected data.Secondly, it takes the key frame technology to solve the requirement of the real-time of the algorithm on the rice combined harvesters. Because the combined harvesters itself has self-adaptability on the rice density during maturation, as long as we detect the image frame which is particularly dense and special thin in rice mature stage, then through automatic control technology it can control the speed of rice combined harvesters. Reduce the forward speed of rice combined harvesters in the special dense rice and increase speed in the special sparse rice in order to maintain feed quantity density of the combined harvester unchanged. The algorithm in the paper which is frame difference based on the histogram is as follows:(1) After the template is certain, then we will extract parameters that is interested in the algorithm within all image frames in the template.(2) Determining the histogram of every frame in the Video Sequence, and comparing with template.(3) Finally, according to the global threshold by the template, and testing, if the absolute value of the template difference value is greater than the threshold, the rice density has changed.This paper adopt four detection algorithms, the average handling time of processing an image is50ms-80ms, and simulation experiments show frame difference based on the histogram takes only50ms to test every frame. The template algorithm solves the idea of algorithm that is unrelated to rice varieties and other environmental factors, and the template algorithm will be about30times faster than the traditional image algorithm(i.e.once per frame.).The algorithm needs to be verified on the rice combined harvesters.
Keywords/Search Tags:Density of rice, Video mining, Shot detection, Feeding volume, Combineharvester
PDF Full Text Request
Related items