TARGET REGION RECOGNITION METHOD OF ENVIRONMENTAL POLLUTION IMAGE BASED ON ITERATIVE CLUSTERING HIERARCHICAL SEGMENTATION

To study the target region recognition method based on iterative clustering hierarchical segmentation of environmental pollution images, it aims to address the overly complex environmental pollution image scenes that has the defect of low level of target region recognition. The method uses median filtering algorithm and linear grayscale change enhancement algorithm to implement noise removal and image enhancement pre-processing for environmental pollution images, and then uses mathematical morphology algorithm and simple linear iterative clustering to segment the target regions of environmental pollution images at two levels of morphology segmentation and depth segmentation to obtain environmental pollution image super pixel units. A density clustering algorithm is used to cluster and fuse the segmented environmental pollution image super pixel units, and the feature matching algorithm is used to compare the location features of the target region of environmental pollution image with the location features of the region to be recognized, and then the target region feature matching recognition results arc output. The experimental results show that the method can accurately identify and locate the target regions of environmental pollution images, and the IoA value of identifying the target regions of environmental pollution images is higher than 0.9.