Because fruit state need to be discriminate to select the different harvesting method before it is fast picked, the fruit state discrimination method was developed. Firstly, the two apple images were acquired, and then segmented by the improved OTSU dynamic threshold segmentation method. Secondly, the picking target fruit was selected based on the principle of the nearest distance to image center when there were a lot of fruits. Thirdly, the two segmented fruit images were done by the inter-frame difference method, then the target fruit state was got using the connection number discrimination to the difference image and the centroid coordinates matching discrimination to the oscillating fruit image. Lastly, the test results showed that the designed algorithm was feasible and effective for most cases in natural environment, and the discrimination time was less than 0.2s.
Deep Learning-Based Congestion Detection at Urban Intersections
