Jinhai Cai

Sensor Network for the Monitoring of Ecosystem: Bird Species Recognition

In this paper, we investigated the performance of bird species recognition using neural networks with different preprocessing methods and different sets of features. Context neural network architecture was designed to embed the dynamic nature of bird songs into inputs. We devised a noise reduction algorithm and effectively applied it to enhance bird species recognition. The performance of the context neural network architecture was comparatively evaluated with linear/mel frequency cepstral coefficients and promising experimental results were achieved.

Hidden Markov models with spectral features for 2D shape recognition

We present a technique using Markov models with spectral features for recognizing 2D shapes. We analyze the properties of Fourier spectral features derived from closed contours of 2D shapes and use these features for 2D pattern recognition. We develop algorithms for reestimating parameters of hidden Markov models. To demonstrate the effectiveness of our models, we have tested our methods on two image databases: hand-tools and unconstrained handwritten numerals. We are able to achieve high recognition rates of 99.4 percent and 96.7 percent without rejection on these two sets of image data, respectively.

A fuzzy Logic Controller for Isolated Signalized Intersection with Traffic Abnormality Considered

This paper presents a fuzzy logic controller for an isolated signalized intersection. The controller controls the traffic light timings and phase sequence to ensure smooth flow of traffic with minimal delay. Usually fuzzy traffic controllers are optimized to maximize traffic flows/minimize traffic delays under typical traffic conditions. Consequentially, these are not the optimal traffic controllers under exceptional traffic cases such as roadblocks and road accidents. We propose a new fuzzy traffic controller that can optimally control traffic flows under both normal and exceptional traffic conditions. In this system, sensors are placed strategically at incoming and outgoing links (lanes) and the controller utilize the information received from these sensors to make optimal decisions to minimize the traffic delays. A simulator is developed to evaluate the performance of traffic controllers under different conditions. Results show that the performance of the proposed traffic controller is similar to that of conventional fuzzy traffic controllers under normal traffic conditions and is better that of others under abnormal traffic conditions.

Evaluation of Aerial Remote Sensing Techniques for Vegetation Management in Power-Line Corridors

This paper presents an evaluation of airborne sensors for use in vegetation management in power-line corridors. Three integral stages in the management process are addressed, including the detection of trees, relative positioning with respect to the nearest power line, and vegetation height estimation. Image data, including multispectral and high resolution, are analyzed along with LiDAR data captured from fixed-wing aircraft. Ground truth data are then used to establish the accuracy and reliability of each sensor, thus providing a quantitative comparison of sensor options. Tree detection was achieved through crown delineation using a pulse-coupled neural network and morphologic reconstruction applied to multispectral imagery. Through testing, it was shown to achieve a detection rate of 96%, while the accuracy in segmenting groups of trees and single trees correctly was shown to be 75%. Relative positioning using LiDAR achieved root-mean-square-error (rmse) values of 1.4 and 2.1 m for cross-track distance and along-track position, respectively, while direct georeferencing achieved rmse of 3.1 m in both instances. The estimation of pole and tree heights measured with LiDAR had rmse values of 0.4 and 0.9 m, respectively, while stereo matching achieved 1.5 and 2.9 m. Overall, a small number of poles were missed with detection rates of 98% and 95% for LiDAR and stereo matching.

RootGraph: a graphic optimization tool for automated image analysis of plant roots

Highlight The method presented analyses root scans automatically, distinguishes primary from lateral roots, and quantifies a broad range of traits for individual primary roots and their associated lateral roots.

Root phenotyping by root tip detection and classification through statistical learning

AimsRoot branching is a fundamental phenotypic property of a root system. In this paper we present a system for the fully automated detection and classification of root tips in root images obtained either by 2D flat bed scanning or by 3D digital camera imaging. With our system we aim to provide a robust, efficient and accurate means of phenotyping of roots.MethodsStructural information derived from image features such as root ends and root branches is utilised for the detection and classification processes. A statistical analysis based on training data sets of root tips and non-root tips is used to assign image features to one of three different classes: non-root tips, primary root tips and lateral root tips. The automated procedure is optimised to ensure as high true detection rate and low false detection rate as possible.ResultsWe apply the method to images of barley, rice, and corn roots taken either by 2D scanning of washed and cut roots or digital camera images of plant roots growing in a transparent medium. The results of our detection and classification procedure are validated by a comparison with manually labelled images for all three species. Our results are also compared to two established platforms, EZ-Rhizo and WinRHIZO. Finally, we demonstrate the utility of the statistical learning approach by quantifying root phenotypic properties of barley double haploid lines.ConclusionsThe method of statistical learning of characteristic features is an accurate means of not only counting root numbers, but also discriminating between primary and lateral roots. The fully-automated procedure presented in this paper can be used reliably in high throughput situations to characterise quantitative phenotypic variation.

Pattern recognition using Markov random field models

In this paper, we propose Markov random field models for pattern recognition, which provide a flexible and natural framework for modelling the interactions between spatially related random variables in their neighbourhood systems. The proposed approach is superior to conventional approaches in many aspects. This paper introduces the concept of states into Markov random filed models, presents a theoretic analysis of the approach, discusses issues of designing neighbourhood system and cliques, and analyses properties of the models. We have applied our method to the recognition of unconstrained handwritten numerals. The experimental results show that the proposed approach can achieve high performance.

High-throughput 3D modelling of plants for phenotypic analysis

In this paper we propose a twin mirror-based system for reconstructing 3D models of real plants for subsequent phenotypic analysis. The method is based on the visual hull concept: multiple reflections of the object from the mirrors give different views of the object and are interpreted as taken from virtual cameras. The epipolar geometry of the object and its four reflections is determined without relying on information of the positions of the camera and mirrors. This alleviates the usual camera calibration step. Two simultaneous images of object mirror scene give ten different and simultaneous views of the plant, without requiring any plant or camera movement. Visual hull algorithms are sensitive to segmentation of the object from the scene. We propose a novel machine learning approach to segment a plant from its background. The plant colours are represented using a Gaussian mixture model (GMM), while the background colours are represented by a separate GMM, learnt using an Expectation Maximisation (EM) algorithm. A Bayes classification rule that satisfies the Neymann-Pearson criteria is used to classify the pixels and thus segment the five plant silhouette from each image. We show results of 3D models of wheat, grass, and a lavender shoot reconstructed using the proposed segmentation and 3D visual hull method.

A new thresholding algorithm based on all-pole model

A new thresholding algorithm based on all-pole model histogram is presented. The parameters of the all-pole model are estimated from the second derivative of log model histogram. The main merits are that the model can produce the desired number of peaks in model histogram and the algorithm is efficient in computation. The proposed algorithm can be used for binarisation and multilevel thresholding and good results have been achieved for a wide range of images.

Genetic diversity for root plasticity and nitrogen uptake in wheat seedlings

Enhancing nitrogen use efficiency (NUE) of wheat is a major focus for wheat breeding programs. NUE may be improved by identifying genotypes that are competitive for nitrogen (N) uptake in early vegetative stages of growth and are able to invest that N in grain. Breeders tend to select high yielding genotypes under conditions of medium to high N supply, but it is not known whether this influences the selection of root plasticity traits or whether, over time, breeders have selected genotypes with higher N uptake efficiency. To address this, genotypes were selected from CIMMYT (1966-1985) and Australian (1999-2007) breeding programs. Genotypes from both programs responded to low N supply by expanding their root surface area through increased total root number and/or length of lateral roots. Australian genotypes were N responsive (accumulated more N under high N than under low N) whereas CIMMYT genotypes were not very N responsive. This could not be explained by differences in N uptake capacity as shown by 15N flux analysis of two representative genotypes with contrasting N accumulation. Expression analysis of nitrate transporter genes revealed that the high-affinity transport system accounted for the majority of root nitrate uptake in wheat seedlings under both low and high N conditions.