Research On Saliency Region Extraction Method Of Leaf Image

Leaves are important organs to maintain the normal growth of trees,which can reflect many important information such as tree species,growth status and so on.Using computer image processing technology to segment the leaf image,the leaf shape,contour,texture and other characteristic information can be obtained by analyzing the leaf image.It has practical significance for tree classification and recognition,resource protection and utilization,exploring the relationship between trees,clarifying the evolutionary law,and the practical application of agriculture and horticulture.This paper analyses the methods of leaf image segmentation currently studied at home and abroad.Aiming at the problem that the leaf image with complex background collected under natural conditions is difficult to get the complete target leaf area which users are most likely interested in directly from the background removed from the image segmentation,it is proposed that the saliency target detection method can be used to obtain the saliency of leaf image directly.Goals.In saliency target detection method,the shortcoming of SLIC super-pixel segmentation method in image preprocessing is firstly improved.The RGB color information of leaf image is transformed into HSV color information.According to the human visual attention resolution ability,the super-image needed by the image is automatically determined by unequal interval quantization according to the subjective color perception and different range of color.Number of prime elements to adapt to the adaptability of computer vision detection applications.Secondly,the super-pixels obtained by SLIC segmentation method are used as nodes to establish absorption Markov chain,and the absorption time from the transfer node to the absorption node is calculated as saliency value to obtain saliency map of the image,which directly extracts the most interesting target region from the background of the leaf image,thus effectively improving the overall detection efficiency of saliency target detection method.To solve the problem of low saliency value in some salient areas of leaf image saliency map,a method based on symmetry region filtering is proposed to segment saliency map of leaf image according to the botanical characteristics of plant leaves with symmetry.Firstly,the symmetry of saliency map is detected to obtain the symmetry axis and center of mass.Then,the saliency map is filtered by the symmetry property to obtain a clear and complete target image segmentation region.The experimental results show that the method is more complete in extracting salient target regions.