Comparison of Signal to Noise Ratio and its Features Variation

摘要

Remote sensing images are representations of ground objects as interpreted from space sensors. Image are captured using reflected electromagnetic signal, on the basis of signal intensity identification of objects are achieved. Thus signal values also depends upon sensor capabilities and object characteristics. Sensor which capture remotely sensed images are analyzed on the basis of ground object spectral values. Using images spectral values, ground object characteristic identification are also achieved. In real scenario noise are comprised along with signal which leads to distracts all objects identifications. It also varies across all over the spectral values and on various feature classes. Signal to noise ratio (SNR) describes the quality of a measurement. Higher signal to noise ratio in the image, spectral values of image helps in identification of ground object of presumable quality. SNR computation is prerequisite process before carrying identification analysis on objects. The SNR represents a useful statistics that are computed and compared across different ground features. Hyperion hyperspectral image data set is used to carry study of Signal to Noise ratio. SNR computation is important process, but it is less studied by researches and scientist communities. SNR are computed using three algorithms Homogenous Area, Nearly Homogenous Area and Geostatistical on various feature classes and compared to evaluate its performance on different features. Geostatistical Algorithms is considering large number of spatial pixels, which are heterogeneous never the less results are varies less in comparison to other used algorithms Homogenous Area and Nearly Homogenous area. Feature Barren land have high SNR while comparing with other feature classes using all three used algorithms. Barren land have high signal reflectance and less absorption by atmosphere. The signal to noise ratio is established to be varying across function of both spectral values and ground features.