Anis Salwa Mohd Khairuddin

Tropical wood species recognition system based on multi-feature extractors and classifiers

An automated wood recognition system is designed to classify tropical wood species. The wood features are extracted based on two feature extractors: Basic Grey Level Aura Matrix (BGLAM) technique and statistical properties of pores distribution (SPPD) technique. Due to the nonlinearity of the tropical wood species separation boundaries, a pre classification stage is proposed which consists of Kmeans clustering and kernel discriminant analysis (KDA). Finally, Linear Discriminant Analysis (LDA) classifier and K-Nearest Neighbour (KNN) are implemented for comparison purposes. The study involves comparison of the system with and without pre classification using KNN classifier and LDA classifier. The results show that the inclusion of the pre classification stage has improved the accuracy of both the LDA and KNN classifiers by more than 12%.

Using Two Stage Classification for Improved Tropical Wood Species Recognition System

An automated wood recognition system is designed based on five stages: data acquisition, pre-processing images, feature extraction, pre classification and classification. The proposed system is able to identify 52 types of wood species based on wood features extracted using Basic Grey Level Aura Matrix (BGLAM) technique and statistical properties of pores distribution (SPPD) technique. The features obtained from both feature extractors are fused together and will determine the classification between the various wood species. In order to enhance the class separability, a pre-classification stage is developed which includes clustering and dimension reduction. K-means clustering is introduced to cluster the 52 wood species. As for dimension reduction, we proposed linear discriminant analysis (LDA) to solve linear data and kernel discriminant analysis/ generalized singular value decomposition (KDA/GSVD) to solve nonlinearly structured data. For final classification, K-Nearest Neighbour (KNN) classifier is implemented to classify the wood species.

Tree species classification based on image analysis using Improved-Basic Gray Level Aura Matrix

The study focuses on classifying wood species based on macroscopic image of wood texture.Use dataset of 52 tropical wood species where 100 images are taken from each wood species.I-BGLAM feature extractor is proposed to extract 136 features from the wood texture.Significant improvement compared to previous system that used GLCM feature extractor. Classifying wood species accurately is crucial since incorrect labelling of wood species may incur huge loss to timber industries. An automated wood species recognition system is designed based on image analysis of the wood texture which consists of image acquisition, feature extraction, and classification. There are 100 images captured from each wood sample which are divided into training samples and testing samples. An effective feature extractor is important to extract most discriminant features from the wood texture in order to distinguish the wood species accurately. Therefore, in this paper, a novel feature extractor based on Improved-Basic Gray Level Aura Matrix (I-BGLAM) technique is proposed to extract 136 features from each wood image. Fundamentally, the proposed I-BGLAM feature extractor which focuses on the gray level of the wood images is rotational invariant and has smaller feature dimension since only discriminative features are considered. Then, the proposed system automatically classifies 52 wood species by using backpropagation neural network classifier. The proposed I-BGLAM feature extractor had shown to overcome the limitations of Gray Level Co-occurrence Matrix (GLCM) and conventional BGLAM feature extractors in wood species recognition system. Experiments were performed to determine which dataset would be the most ideal when dividing the 100 wood images into training samples and testing samples. Results showed that the most ideal dataset that should be used is dataset that consists of 80 training samples and 20 test samples. The proposed method showed marked improvement of 97.01% accuracy to the work done previously.

Automatic online signature verification: A prototype using neural networks

Signature verification is the process used to recognize an individuals handwritten signature to prevent fraud. In this paper pressure at the pen-tip together with the x, and y coordinates of the signature are measured and features extracted from these are used to verify the signature. A pressure pad was used to obtain signature samples. A signature verification system using SOM neural network was designed in MATLAB to verify the signatures. Results obtained using a prototype system are encouraging. The attractive features of this system are its low cost, low intrusion, good performance and use of an acceptable and natural biometric (the signature).

Tree species recognition system based on macroscopic image analysis

Abstract An automated wood texture recognition system of 48 tropical wood species is presented. For each wood species, 100 macroscopic texture images are captured from different timber logs where 70 images are used for training while 30 images are used for testing. In this work, a fuzzy pre-classifier is used to complement a set of support vector machines (SVM) to manage the large wood database and classify the wood species efficiently. Given a test image, a set of texture pore features is extracted from the image and used as inputs to a fuzzy pre-classifier which assigns it to one of the four broad categories. Then, another set of texture features is extracted from the image and used with the SVM dedicated to the selected category to further classify the test image to a particular wood species. The advantage of dividing the database into four smaller databases is that when a new wood species is added into the system, only the SVM classifier of one of the four databases needs to be retrained instead of those of the entire database. This shortens the training time and emulates the experts’ reasoning when expanding the wood database. The results show that the proposed model is more robust as the size of wood database is increased.

Denoising module for wood texture images

The need for an effective automatic wood species identification system is becoming critical in the timber industry with the intention to sustain and improve productivity and quality of the timber products in furniture industry and housing industry. The first stage in an automatic wood recognition system is the image acquisition process where wood images are captured and stored in the database. Good quality wood images must be obtained during the acquisition process in order to guarantee effective results. One of the main issues in identifying wood species effectively is the blurred images of wood texture captured during the image acquisition process. To cater the above-mentioned problem, wood image denoising process is crucial for the timber industry. An image denoising module is proposed to improve the image representation of the wood texture by using the expectation–maximization (EM) adaption algorithm. Then, image quality assessment techniques are applied to evaluate the quality of the denoised wood images. Finally, the performance of the proposed denoising technique is compared to several denoising techniques at various noise levels. In this research, 52 wood species are used where the size of each wood image is 768 × 576 pixels with 256 gray levels at 300 dpi resolution. Experimental results tabulate the mean and standard deviation of the image quality assessment values for each technique at various noise levels. It can be seen that the proposed method EM adaption filter gives the best peak signal-to-noise ratio performance compared to other techniques. In conclusion, the proposed EM adaptation method gives the best performance in denoising the wood texture images at various noise levels compared to other techniques, such as homomorphic filtering, direct inverse filter, Wiener filter, constrained least squares, Lucy–Richardson algorithm, and EM filter.

Waste level detection and HMM based collection scheduling of multiple bins

In this paper, an image-based waste collection scheduling involving a node with three waste bins is considered. First, the system locates the three bins and determines the waste level of each bin using four Laws Masks and a set of Support Vector Machine (SVM) classifiers. Next, a Hidden Markov Model (HMM) is used to decide on the number of days remaining before waste is collected from the node. This decision is based on the HMM’s previous state and current observations. The HMM waste collection scheduling seeks to maximize the number of days between collection visits while preventing waste contamination due to late collection. The proposed system was trained using 100 training images and then tested on 100 test images. Each test image contains three bins that might be shifted, rotated, occluded or toppled over. The upright bins could be empty, partially full or full of garbage of various shapes and sizes. The method achieves bin detection, waste level classification and collection day scheduling rates of 100%, 99.8% and 100% respectively.

Statistical feature extraction method for wood species recognition system

A cascaded wood species recognition system using simple statistical properties of the wood texture is presented where a total of 24 statistical features are extracted from each wood sample. They are mainly vessel features that allow a broad initial grouping of wood texture using fuzzy logic. Then, a neural network classifier is used to refine the broad grouping into the final wood species classification. The proposed system emulates the classification approach normally taken by human experts when analyzing wood species based on texture. A comprehensive set of experiments was performed on a database composed of 3000 macroscopic images of 30 different wood species to evaluate the effectiveness of the system. Finally, its performance is compared with previous works in terms of classification accuracy.

Wood Species Recognition System based on Improved Basic Grey Level Aura Matrix as feature extractor

An automated wood species recognition system is designed to perform wood inspection at custom checkpoints in order to avoid illegal logging. The system that includes image acquisition, feature extraction and classification is able to classify the 52 wood species. There are 100 images taken from the each wood species is then divided into training and testing samples for classification. In order to differentiate the wood species precisely, an effective feature extractor is necessary to extract the most distinguished features from the wood surface. In this research, an Improved Basic Grey Level Aura Matrix (I-BGLAM) technique is proposed to extract 136 features from the wood image. The technique has smaller feature dimension and is rotational invariant due to the considered significant feature extract from the wood image. Support vector machine (SVM) is used to classify the wood species. The proposed system shows good classification accuracy compared to previous works.