Exploring Illumination Robust Descriptors for Human Epithelial Type 2 Cell Classification
Xianbiao Qi, Guoying Zhao, Jie Chen, Matti Pietikainen,
Oulu University, Finland
Abstract
Strong illumination variation is a key challenge in the Human Epithelial Type 2 (HEp-2) cell classification task. Aiming to improve the robustness of the HEp-2 classification system to the illumination variation, this paper deeply explores discriminative and illumination robust descriptors. To be more specific, we propose a novel Spatial Shape Index Descriptor (SSID) to capture spatial layout information of the second-order structures, and utilize a Local Orientation Adaptive Descriptor (LOAD), which was originally designed for texture classification, to the HEp-2 cell classification task. Both SSID and LOAD show strong discrimination. Meanwhile, both features show great complementarity to each other. Four different sets of experiments are carried out to evaluate the SSID, the LOAD and their combination. Note that, two submissions both achieved superior performance on the new Executable Thematic of Pattern Recognition Techniques for Indirect Immunofluorescence images analysis. Compared to the first place method of the ICPR 2014 HEp-2 cell classification contest that combined four types of features, both of our submissions outperformed it with only one type of feature. Meanwhile, we evaluated our features on a newly created large-scale HEp-2 data set, and the superior performance achieved on this data set further confirmed the effectiveness of our features.
Paper
Xianbiao Qi, Guoying Zhao, Jie Chen and Matti Pietikainen. Exploring Illumination Robust Descriptors for Human Epithelial Type 2 Cell Classification. Submitted to Pattern Recognition.
Our Results
Source Code
LOAD Evaluation SSID Evaluation Shape Index Histogram Evaluation
Here, we compare our LOAD and SSID with the state-of-the-art descriptor Shape Index Histogram(SIH), if you use the provided SIH code, please cite the reference [1].Our Newly Created Dataset
References