Fast Plant Leaf Recognition Using Improved Multiscale Triangle Representation and KNN for Optimization

Abstract

Due to the complexity and similarity of plant leaves, it is very important to study an effective leaf-feature extraction method to improve the recognition rate of plant leaves. We study five multiscale triangle representations: the triangle unsigned area representation (TUA), the triangle vertex angle representation (TVA) and three new representations, which we define as the gray average (TGA), the gray standard deviation (TGSD) and the side length integral (TSLI) on the triangle. In this method the curvature features of the contour, the texture features and the shape area feature are extracted to provide a multiscale leaf-feature description, and a new adaptive KNN for optimization method is proposed to improve the retrieval rate of leaf datasets. Experiments show that compared with the state-of-the-art methods, our method has higher accuracy on the Swedish and Flavia plant leaf datasets, which are respectively 99.35% and 99.43% with 84.76% Mean Average Precision (MAP) value and has comparable results on MPEG-7, kimia99 and kimia216 datasets. When our method is combined with KNN for optimization, the retrieval rate of the above datasets has been significantly improved, especially MAP on the Flavia dataset increases to 94.48%.INDEX TERMS Plant leaf recognition, multi-scale leaf-feature description, multi-scale triangle representation, adaptive KNN for optimization.

Leaf Recognition Using Convolutional Neural Networks Based Features

“…The performances were almost the same, 99.74% ± 0.42% on validation sets and 99.58%±0.31% on test sets. Until now, the best recognition rate for the Flavia dataset was 99.43% [21]. Thus, our experimental results show that the proposed approach could bypass the current state-of-the-art methods.…”

“…Given the Faliva dataset, Kumar et al [13] extracted shape and edge attributes before putting these features into the KNN model, and then they could obtain an average precision of 94.37%. In 2020, Su et al [21] proposed five multi-scale triangle representations for each leaf image after extracting the leaf's form to improve the classification performance. This approach used KNN as an optimization method and could achieve outstanding performance with an accuracy of 99.43%.…”

“…Nevertheless, in terms of accuracy, if computer vision techniques can tackle these limitations before extracting engineered features, the performance is approximately equal to that of neural networks. Indeed, the highest performance of using purely handcrafted attributes so far is 99.43% [21] while that of deep learning techniques is 99.42% [24] in the Flavia dataset. Multiple studies have employed both hierarchies for this plant recognition taskconventional machine learning algorithms and neural networks.…”

Fast Target Recognition Based on Improved ORB Feature

“…The core idea of the K-nearest-neighbor (KNN) algorithm [34][35][36][37][38][39][40][41] is to search the top K feature points with the highest similarity in another feature space as candidate features. After adaptive histogram equalization, the details of the dark part of the original image become clearer and the candidate points with a higher Harris response value change during feature point detection, so high-quality matching point pairs in other areas of the same image can be obtained.…”

“…The core idea of the K-nearest-neighbor (KNN) algorithm [34][35][36][37][38][39][40][41] is to search the top K feature points with the highest similarity in another feature space as candidate features.…”