Statistical Texture Analysis of Microscopy Images for the Purpose of Surface Characterization
- PDF / 313,036 Bytes
- 6 Pages / 612 x 792 pts (letter) Page_size
- 56 Downloads / 136 Views
0894-LL03-06.1
Statistical Texture Analysis of Microscopy Images for the Purpose of Surface Characterization T. E. Sadowski*†, C. Caragianis-Broadbridge*, J. Daponte† *Department of Physics, Southern Connecticut State Univ., New Haven, CT 06515 †Department of Computer Science, Southern Connecticut State Univ., New Haven, CT 06515 ABSTRACT This study focuses on the application of an image processing technique known as texture analysis to provide a complementary method to process non-contact atomic force microscopy data. For the current study, SrBi2Ta9O9 (SBT) thin films (~200nm) on silicon samples were prepared via MOD and imaged via nc-AFM. Previous studies have qualitatively chosen grain size or grain density to quantify the surface characteristics of SBT but for this study texture analysis was used. Results obtained from the texture analysis and the subsequent stepwise discriminant analysis illustrated the importance of grain density and its effect on the statistics obtained from the gray scale cooccurrence matrix. INTRODUCTION As personal computing devices continue to become smaller and faster, there is an increased interest in the study of new materials that will allow for continued innovation. One such material is the ferroelectric (FE) thin film SrBi2Ta9O9 (SBT) which has possible applications as a new type of non-volatile computer memory. Before these materials can be used for such purposes, their microscopic physical structure must be characterized so that a relationship can be determined between processing, structure, and the resulting material characteristics. Often this requires the use of a myriad of complementary imaging modalities such as transmission electron microscopy (TEM), scanning electron microscopy (SEM), and atomic force microscopy (AFM). Although these techniques yield data regarding the structure of a given material, they can prove to be quite labor intensive if multiple images need to be evaluated. The focus of this research is the use of an imaging processing technique known as texture analysis to provide a complementary method to analyze images acquired from thin film SBT samples using non-contact atomic force microscopy (nc-AFM). Previous studies have indicated a strong correlation between the surface topology of SBT thin films at a particular annealing temperature and the behavior of the resulting device [1]. Texture, one of the most important characteristics used to identify objects or regions of interest in an image, has previously proved useful in automation of the inspection process in the lumber, leather, and textiles industries [2]. Although there are many approaches toward texture analysis, it is most often quantified with a set of gray scale cooccurrence matrices used to obtain what are commonly referred to as second order statistics about pixel values in an image. A cooccurrence matrix is a matrix of relative frequencies Pi j ,
where each element (i, j ) is the sum of the number of times a pixel with gray level i occurs at a distance d from a pixel with gray level j along an
Data Loading...
