The Analysis of Image Texture Feature Robustness Using CometCloud

The robustness of image features is a very important characteristic for quantitative image analysis. The objective of this paper is to investigate the robustness of various texture features using Hematoxylinstained breast tissue microarray slide by simulating different practical imaging problems including out of focus, magnification changes, illumination variations, noise, compression, distortion and rotation due to the process of preparing the specimens. We employed five texture analysis methods and tested with all these challenges listed above. Our evaluated texture features include co-occurrence matrix, center-symmetric autocorrelation, texture feature coding method, local binary pattern, and texton. Due to the independency of each transformation and texture descriptor, a network structured combination is proposed and deployed on the Rutgers private cloud. The whole process is evaluated using 20 randomly selected tissue microarray cores. All the combinations of the image transformations and deformations are calculated, and the whole feature extraction procedure was completed in 70 minutes using a cloud with 20 nodes. Center-symmetric auto-correlation outperforms all the other four texture descriptors but requires the longest computational time. It is roughly 10 times slower than local binary pattern and texton. On the basis of speed considerations, both local binary pattern and texton provided excellent performance for classification and content based image retrieval.