This paper presents a dynamic classifier selection approach for hyperspectral image classification, in which both spatial and spectral information are used to determine a pixel’s label once the remaining classified pixels’ neighborhood meets the threshold. For volumetric texture feature extraction, a volumetric gray level co-occurrence matrix is used; for spectral feature extraction, a minimum estimated abundance covariance-based band selection is used. Two hyperspectral remote sensing datasets, HYDICE Washington DC Mall and AVIRIS Indian Pines, are employed to evaluate the performance of the developed method. The classification accuracies of the two datasets are improved by 1.13% and 4.47%, respectively, compared with the traditional algorithms using spectral information. The experimental results demonstrate that the integration of spectral information with volumetric textural features can improve the classification performance for hyperspectral images.
We present a fast, simple, sub-pixel algorithm on the critical angle refractometer to measure the refractive index of the liquid sample by determining the centroid of the light intensity of the relative reflective curve. The centroid algorithm utilizes a divergent fiber-coupled royal blue LED source to irradiate on the dielectric surface between the prism and the media, which generates the light intensity distribution of the reflectance facula. Instead of the critical angle pixel as the differential algorithm and the threshold algorithm, the sub pixel centroid algorithm is based on calculating the centroid value of the light intensity of the relative reflective curve. In some moderate turbid solutions, the centroid algorithm is less sensitive to the scattering and absorption than the differential algorithm and the threshold algorithm. It is possible to utilize the centroid point of the relative reflective curve to determine the refractive index. Supported by the theoretical analysis and experimental results on saline solutions, we can conclude that the proposed algorithm is effective to get the super resolution and meaningful to the refractive index measurement of the liquid. The critical angle refractometer with this centroid method is potential to be a high-accuracy, high-resolution, and reliable automatic refractometer.
An improved calibration method for digital Abbe refractometer is proposed. Based on Fresnel reflection theory, digital
Abbe refractometer measures the index of refraction by processing bright-dark pattern images. For extreme environment
applications, our team has developed a digital Abbe refractometer. By analyzing bright-dark pattern images, it shows
optical aberration may reduce positioning accuracy on critical angle. The main work of this paper is to propose a new
calibrate method for digital Abbe refractometer. An optical system is built to simulate the refractometer. A motorized
micropositioning stage is inserted to precisely control the position of bright-dark boundary. An area CCD captures an
image each time boundary displaced. Get the boundary through entire measuring range to form an image database. The
database indicates the corresponding relations between bright-dark pattern image acquired by CCD camera and boundary
position read by motorized stage. When measuring the refractive index of liquid, match its bright-dark pattern image to
images in database, and get the boundary position from the nearest match. Compared to the former method of computing
boundary position from images with aberration, the proposed method calibrate refractometer by large amount of
experimental data, thus improve stability of the measurement.