Plant leaf distinguishing method based on image segmentation and feature fusion

Abstract

The invention relates to a plant leaf distinguishing method based on image segmentation and feature fusion. According to the method, the plant leaf identification in a complex background is divided into two stages of leaf segmentation and feature extraction. The morphological transform-based marker-driven watershed algorithm is mainly adopted for starting and stopping the reconstructing operationon plant leaves. After that, plant leaves are segmented by adopting the marker-driven watershed algorithm. As a result, the over-segmentation phenomenon caused by the influence of leaf textures and veins during the segmentation process of plant leaves is avoided. Therefore, the negative effect of the complex background factor on plant leaf identification is eliminated.

Description

technical field [0001] The invention relates to a complex background plant recognition method based on leaf image segmentation and feature fusion. The method intends to use the marking watershed algorithm based on morphological transformation to solve the problem of plant recognition and plant leaf segmentation in complex backgrounds through leaf segmentation and feature extraction. The method solves the over-segmentation problem in the process, thereby eliminating the negative impact of complex background factors on plant identification, and belongs to the technical field of plant identification. Background technique [0002] There are many kinds of plants on the earth, and there are large errors in identifying plants through naked eye observation. Therefore, plant classification and identification based on computer vision has become a research hotspot in plant taxonomy. As an integral part of plants, plant leaves are often used as important features for plant identificatio...

AI Technical Summary

Problems solved by technology

[0003] Aiming at the processing of leaves under complex backgrounds, there is a method of (1) first Gustafson-Kessel clustering and segmentation, and then using genetic algorithm to extract leaves, but the algorithm is less robust to overlapping and complex backgrounds

(2) The watershed algorithm of automatic marking segmented the leaf image, and then extracted Hu moments and Zernike moments as leaf shape features for the leaves to identify the plants. Although this method can better segment leaves under overlapping and complex backgrounds, but The leaf shape information is needed for segmentation, which is not practical for unknown leaves

However, due to the existence of leaf texture and leaf veins, this algorithm is not applicable to most leaves. When the segmentation object has obvious veins, it is easy to cause over-segmentation.

Images