Retraction Note: Agricultural super green image segmentation method based on Gaussian mixture model combined with Camshift

Abstract

The Gaussian mixture model is used to determine the coordinate distribution curve of the image gray value. The distribution probability value and approximate spatial coordinates of the image gray value in different intervals are calculated by substituting the specified data into the model. When the approximate spatial coordinates are determined, establishing a second-order entropy function is used to calculate the coordinate distribution set of different levels of gray in a specific range, and the gray with the most considerable entropy function is selected as the distinction among all grays. The objectives are proved through a large number of simulation experiments that the method of establishing an entropy function to determine the spatial distribution of image gray levels can improve the accuracy of image segmentation and enhance image processing efficiency. The Camshift algorithm is frequently used when using the entropy function to determine the spatial distribution of image gray levels. At present, the algorithm is widely used in image processing and widely used by researchers in the research of computer data processing and robot target tracking. However, this algorithm has a fatal flaw. Suppose the color of the environment where the target is tracked is very similar to the color of the target itself or is prone to interference. In that case, it is straightforward to lose the target during target tracking. Researchers put forward a target tracking method based on ORB feature detection and Kalman filtering multiple algorithms based on this phenomenon. First, detect the target ORB feature points to initialize the search window, and then use the Kalman filter as the target motion state prediction mechanism. Suppose it is interfered with by the background. In that case, the ORB algorithm is used to match the Kalman prediction area’s feature points and the target model in the current frame, and the position of the target in the video frame is detected again. The predictor parameters are updated according to the Kalman filter to predict the target’s possible position after being occluded by the object.