Research Of Non-destructive Grading Technology And Detection Equipment For Hami Big Jujubes Based On Computer Vision

Hami big jujubes is a characteristics fruit and the geography trademark products in Xinjiang of China,to achieve the brand strategy of green food must be detected and graded for harvested Hami big jujubes.Atpresent the detection and grading of Hami big jujubes is still by manual，but manunal work mainly exist theshortcoming such as large amount of labor, low production efficiency, accuracy of grading is unstable, thegrading standards are easily affected by subjective factors of human, it’s difficult to ensure the foodhygiene and security when the workers contact with Hami big jujubes. Existing mechanical sortingmachine has improved the grading efficient, while the grading is accorded to the size, the grading standardis single, serious damage and difficult to improve the quality of jujubes. So need a system which can berealized automatic and non-destruction grading, to improve the grading standard and reduce the cost ofgrading. This paper is mainly to detect the external quality of dried Hami big jujubes based on computervision technology, achieve to sort efficiently and quickly, ensure the quality of supply and export ishigh-quality Hami big jujubes, at the same time to improve the price. This paper has mainly completed thefollowing contents:(1) Obtained images of the test samples using the existing and non-destructive test-bed for Hamibig jujubes, determined the background and the arrangement of light source according to the characteristicsof color and fruit surface. Pretreated the images by MATLAB, ascertained and optimized the algorithm,select （0.1,0.7）for Hami big jujubes by threshold segmentation method，dislodge the images’ backgroundof Hami big jujubes by color space conversion.(2) Introduced the frequently used edge detection operators and application of wavelet transform inthe image edge detection, analyzed the results of image edge detection to find the best way to detect thewrinkle of Hami big jujubes.(3) The wrinkles of Hami big jujubes as one of the important standards of grading, decomposed theimages of removing the background using the wavelet，as can be seen from the decomposed images that thelow-frequency showed the outline of images, the high-frequency showed the details of images, extractedthe decomposed images’ wavelet coefficient, fused the images, then obtain the images that the surfacewrinkles is obvious, extracted the characteristics of surface wrinkles base on the computer visiontechnology.(4) By measuring the size and weight of Hami big jujubes to design automatic feeding device,realized to feed single by single, PVC belt was used as transmission device, the installation dimension ofbaffles according to the size of Hami big jujubes, adjusted the transmission velocity according to the timeof obtain images by computer vision system, realized to feed single by single, for the realization ofdetecting the external quality of Hami big jujubes, the Hami big jujubes need to flip when they passthrough the region of image acquiring, in order to get their full surface image. Designed the transmissiondevice according to the size and weight of Hami big jujubes, analyzed and simulated the force and motionof Hami big jujubes，got the relationship of Hami big jujubes’ angular velocity and rollers’ velocity.