Blood Cell Image Analysis

A two-phase framework for white blood cell identification. Preprocessed microscopic smears via 2D Butterworth low-pass filtering and LAB color-space segmentation. Fused deep convolutional embeddings (EfficientNetB0, ResNet50, VGG) with handcrafted texture metrics (LBP, GLCM), optimized with Principal Component Analysis (PCA) retaining 95% variance, and classified with SVM.