Hybrid Feature Extraction based on HOGHT to Detect Tumor in Mammogram Images
M. Poomani Alias Punitha, K. Perumal · 2019
The Mammogram image segmentation and feature extraction techniques still need improvement in tumor detection and play a vital process of diagnosing. This paper optimizes the novel approaches of hybrid feature extraction technique for mammogram tumor detection with implicit knowledge. The main objective of this paper is to apply the double threshold segmentation (DTS) with a new hybrid feature extraction called Histogram oriented gradients with Hough transform (HOGHT) and PCA Algorithm. This Extraction can be classified in to benign, normal and malignant. This approach consists of a better result and reduces the Computation time. Quality measure can be done by using this predictive accuracy and no one can be applied for this combination of work. This technique ensures that decision is taken ahead of tumor detection rather than the existing system. The experiments have been made from the dataset taken from MIAS. The obtained results by using this method are highly encouraged.