The model divides the three-dimension space into twenty-seven regions by extending the edges of Minimum Bounding Rectangular(MBR) of reference object and uses a three rows and nine ranks matrix which is constructed by recording whether the regions meet with primary object in the twenty-seven regions to represent direction relations.

In order to enhance the operation speed of the algorithm,the algorithm to calculate the least surrounding boxes was improved.

The problem was discussed that a minimum containment rectangle is transformed to function optimization and meanwhile the optimized solution was found with genetic algorithm.

To seek a convenient method for detection of rice shape,a real-time vision-based detection approach was developed,which used an improved Ostu algorithm to segmentate the image,eliminated the noise of image by open operation,and calculated rice shape characteristic values by the minimum enclosing rectangle method.

A method to inspect the ratio of rice kernel was proposed, in which a calculation method by using the minimum enclosing rectangle (MER) around the rice kernel was designed and verified.

Then this paper proposed a user-defined template based on the minimum enclosing rectangle(MER) to complete the finally orientation,and found the center of the eyes using the area.