Narrow Results

In this paper we assess the effect that clustering pixels into spectrally-similar background types, for example, soil, vegetation, and water in hyperspectral visible/near-IR/SWIR imagery, prior to applying a detection methodology has on material detection statistics. Specifically, we examine the effects of data segmentation on two statistically-based detection metrics, the Subspace Generalized Likelihood Ratio Test (Subspace GLRT) and the Adaptive Cosine Estimator (ACE), applied to a publicly-available AVIRIS datacube augmented with a synthetic material spectrum in selected pixels. The use of synthetic spectrum-augmented data enables quantitative comparison of Subspace-GLRT and ACE using Receiver Operating Characteristic (ROC) curves. For all cases investigated, Receiver Operating Characteristic (ROC) curves generated using ACE were as good as or superior to those generated using Subspace-GLRT. The favorability of ACE over Subspace-GLRT was more pronounced as the synthetic spectrum mixing fraction decreased. For probabilities of detection in the range of 50-80%, segmentation reduced the probability of false alarm by a factor of 3–5 when using ACE. In contrast, segmentation had no apparent effect on detection statistics using Subspace-GLRT, in this example.

In order to solve the problems of low efficiency and long running time caused by the traditional Zernike moment method for convolution calculation of the whole image, this paper combines the canny detection algorithm with the Zernike moment method. First, the canny edge detection algorithm, which combined with the Otsu threshold method, is used to extract the pixel edge of the image. Then an improved Hough transform method is used to fit the geometric edge in the image. Based on this, the Zernike moment method is applied to realize sub-pixel positioning of images. The algorithm improves the deficiencies of direct sub-pixel detection, improving accuracy and reducing running time. To verify the effectiveness of the proposed algorithm, the algorithm is applied to the dimension measurement experiment of T-type guide way. The results clearly show that the algorithm is superior to the traditional algorithm in accuracy.

There are many pixel-level and sub-pixel-level edge localization methods, but some methods perform poorly in the presence of noise and require image pre-processing or iterative methods to improve the localization accuracy, which increases the computational cost. In this paper, we propose an improved Zernike moment subpixel edge localization algorithm based on the ramp model, and combine the idea of bilinear interpolation method to optimize the selection of parameters in the edge model. Through the validation of several examples, it is found that the algorithm outperforms the compared methods for edge localization of noisy images. After the edge detection, in order to improve the measurement accuracy of polyvinyl chloride (PVC) plates, a parallel line fitting method is proposed to fit the edge points, thus avoiding the interference of extraneous noise points and achieving accurate measurement of PVC plate size. The experiments were carried out for several measurements of the sheet length, and the method was verified to have high measurement accuracy.

In this paper we assess the effect that clustering pixels into spectrally-similar background types, for example, soil, vegetation, and water in hyperspectral visible/near-IR/SWIR imagery, prior to applying a detection methodology has on material detection statistics. Specifically, we examine the effects of data segmentation on two statistically-based detection metrics, the Subspace Generalized Likelihood Ratio Test (Subspace GLRT) and the Adaptive Cosine Estimator (ACE), applied to a publicly-available AVIRIS datacube augmented with a synthetic material spectrum in selected pixels. The use of synthetic spectrum-augmented data enables quantitative comparison of Subspace-GLRT and ACE using Receiver Operating Characteristic (ROC) curves. For all cases investigated, Receiver Operating Characteristic (ROC) curves generated using ACE were as good as or superior to those generated using Subspace-GLRT. The favorability of ACE over Subspace-GLRT was more pronounced as the synthetic spectrum mixing fraction decreased. For probabilities of detection in the range of 50-80%, segmentation reduced the probability of false alarm by a factor of 3–5 when using ACE. In contrast, segmentation had no apparent effect on detection statistics using Subspace-GLRT, in this example.