Automated hair removal in dermoscopy images using shallow and deep learning neural architectures

•Processing for hair removal is fundamental for dermoscopy image analysis.•Hair segmentation by multilayer perceptron with filter-based pixel-wise features.•Hair segmentation with end-to-end Encoder-Decoder.•Results are given in terms of hair-free image quality and accuracy of image classification.•Hair removal from dermoscopy images improves melanoma classification versus lesions.