Rapid measurement for 3D geometric features of maize ear based on image processing

At present, the 3D geometric features of maize ear (such as ear volume, ear shape, etc) are usually measured by hand during the breeding process, which needs a lot of workforce. In addition, the subjective errors are not easy to be avoided and the efficiency of manual measurement is very low. In this paper, a rapid measurement method for 3D breeding feature of multiple maize ears is realized based on image HSV color space. Firstly, the exact contours of multiple maize ears are extracted rapidly based on the HSV histogram threshold segmentation algorithm. And the segmentation algorithm is suitable for arbitrarily placed maize ears. Secondly, based on the obtained minimum exterior rectangle of ear contour and cumulative sum procedure of contour sections, the rapid and accurate measurement of three dimensional geometric feature of maize ear is realized. Finally, taking the ZD958 and NH101 for examples, the above algorithm is verified. The measurement accuracy of ear volume can reach to 94%. The error causes are also analyzed further. The algorithm proposed in this paper can measure the 3D breeding features of multiple maize ears rapidly and greatly improve the evaluation efficiency of maize varieties.