Automatic Shadow Detection in Aerial and Terrestrial Images

Abstract

Abstract:Shadows exist in almost all aerial and outdoor images, and they can be useful for estimating Sun position estimation or measuring object size. On the other hand, they represent a problem in processes such as object detection/recognition, image matching, etc., because they may be confused with dark objects and change the image radiometric properties. We address this problem on aerial and outdoor color images in this work. We use a filter to find low intensities as a first step. For outdoor color images, we analyze spectrum ratio properties to refine the detection, and the results are assessed with a dataset containing ground truth. For the aerial case we validate the detections depending of the hue component of pixels. This stage takes into account that, in deep shadows, most pixels have blue or violet wavelengths because of an atmospheric scattering effect.Keywords: Shadow Detection; Aerial Images; Terrestrial Images. Resumo:Sombras estão presentes na maior parte das imagens aéreas e terrestres, e elas podem ser úteis para estimação da posição do Sol, ou para medir os tamanhos de objetos. Por outro lado, elas representam um problema em processamentos tais como detecção/reconhecimento de objetos, correspondência entre imagens, etc., pois podem ser confundidas com objetos escuros e mudar as propriedades radiométricas da imagem. Neste trabalho esse problema foi tratado em imagens aéreas e terrestres. Utilizou-se um filtro para encontrar áreas com baixas intensidades. Para as imagens terrestres, foram analisadas propriedades da razão do espectro das imagens para refinar a detecção, e os resultados foram avaliados por meio de um conjunto de dados contendo a verdade de campo. Para o caso aéreo as detecções são validadas dependendo da componente da matiz dos pixels. Esse estágio leva em consideração que em sombras profundas a maior parte dos pixels possuem comprimentos de onda na região do azul e violeta por conta de um efeito de espalhamento atmosférico.

A Backpack-Mounted Omnidirectional Camera with Off-the-Shelf Navigation Sensors for Mobile Terrestrial Mapping: Development and Forest Application

“…Forest, especially the dense one, has a huge illumination variation and shadow occurrence in the images. The presence of shadows can negatively affect image quality and photogrammetric processes, including image matching, image classification, automatic detection of objects and loss of information, which requires an image pre-processing step [ 38 ]. Therefore, it is also important to highlight the possibility of combining an omnidirectional system with independent georeferencing capability with other complementary technologies, such as a LASER scanner.…”

Color-based shadow detection method in aerial images

“…They claimed that wavelengths near blue light and the longest near-infrared were good choices to detect shadows in agricultural fields. Colorbased shadow detection algorithm proposed in [11] utilized intensity and saturation elements from the HIS color model. Meanwhile, Agarwal et al [12] proposed an algorithm based on the spatial distribution of neighboring pixels to detect shadows in drone images.…”

Shadow Detection in Hyperspectral Images Acquired by Uav

“…Shadow detection algorithm in color aerial images was also developed by Freitas et al, 2017 andSantos et al (2006). Freitas et al (2017) developed an algorithm based on the hypothesis that most deep shadows have blue and violet wavelengths (Adler-Golden et al, 2002;Polidorio et al, 2003).…”

“…Shadow detection algorithm in color aerial images was also developed by Freitas et al, 2017 andSantos et al (2006). Freitas et al (2017) developed an algorithm based on the hypothesis that most deep shadows have blue and violet wavelengths (Adler-Golden et al, 2002;Polidorio et al, 2003). They computed an index previously proposed by Polidorio et al (2003), which is computed using the intensity and saturation components from HIS color space in order to take advantage of the stronger scattering in the violet and blue wavelengths.…”

“…Shadow pixels have low intensity but high saturation of the visible scattered light. Freitas et al (2017) combined that index with a shadow filter as proposed by Santos et al (2006) in order to detect shaded pixels.…”