técnicas de processamento de imagem : Uma revisão Dust detection in solar panel using image processing techniques : A review

The performance of a photovoltaic panel is affected by its orientation and angular inclination with the horizontal plane. This occurs because these two parameters alter the amount of solar energy received by the surface of the photovoltaic panel. There are also environmental factors that affect energy production, one example is the dust. Dust particles accumulated on the surface of the panel reduce the arrival of light to the solar modules, reducing the amount of generated energy. The cleaning or mitigation of the modules is important and, to optimize these processes, constant monitoring and evaluation must be carried out. In order to increase the efficiency of photovoltaic panels, the use of image processing methods can be considered for the detection of dust. Therefore, the creation of a document that gathers and analyzes the results of different works developed to solve this problem facilitates access to information, allowing a better understanding of what has already been done and how it can be improved. The objective of this article is to review researches that uses image processing techniques to detect dust on solar panels, in order to compile information to assist research in the area and provide inspiration for future studies.

To improve the performance of solar panels, researchers seek to develop methods for detecting surface dust. This can make maintenance work cheaper and simpler, reducing the use of resources and personnel, allowing energy generation with the minimum possible loss. Therefore, this article aims to evaluate research related to image processing applied to the detection of dust in solar panels. This work evaluates, in a condensed form, the literature already produced, reinforcing that these techniques help to improve the performance and useful life of solar panels. Despite the existence of existing research indicating the negative impacts caused by dust in the production of solar energy, with presentation of techniques to mitigate this problem, there is no research presenting the results of image processing techniques in the detection of dust.
The work is organized as follows: section 2 is the theoretical background of the solar energy generation, the effects of dust and a brief introduction on image processing. Section 3 presents the research methodology in this review. Section 4 analyzes the results and techniques used by the selected works. In section 5, discusses the advantages and disadvantages of these image processing applications. In addition, the authors presents some possible improvement solutions for future works. Section 6 is the conclusion, pointing out the knowledge that can be obtained through this review.

Solar panel
Photovoltaic technology uses solar cells built with semiconductors to absorb sun's radiation, converting it into energy. Currently, solar energy plays a prominent role in the production of clean and sustainable energy. However, research focused on the use of semiconductors showed limitations in the efficiency of photovoltaic systems of 15 to 20%.
Thus, research was conducted to improve the efficiency of photovoltaic systems, which resulted in improvements with the application of solar trackers and algorithms for both  Development, v. 9, n. 8, e3219885107, 2020 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v9i8.5107 5 The characteristics of a photovoltaic module can be demonstrated by power-voltage curves. Figure 1 shows the power-voltage curve of a photovoltaic module for different conditions of irradiance and cell temperature. There are factors that can be mitigated, such as dust, for example. However, climatic factors preclude any intervention in a completely satisfactory manner. The technology advances to mitigate problems related to dust, with automatic supervision of solar panels through computer vision and cleaning them if necessary using robots. In addition, for climate problems, only partial solutions exist. An example is an automated system that track solar movement, but it can do nothing about a totally cloudy day.

Dust effect
A variable that can negatively affect the production of energy in a solar panel is the accumulation of dust on it. Dust alters the energy production of photovoltaic modules in two ways. First, the dust particles suspended in the atmosphere, which are longer than the wavelength of the beam of sunlight, scatter the light, reducing the amount of radiation  To perform this research, it was used the following search algorithm: ("dust" OR "soil") AND ("deposition" OR "accumulation") AND ("detection" OR "inspection") AND ("solar" OR "photovoltaic" OR "pv") AND ("panel*" OR "farm*" OR "plant*") AND (1). After removing the duplicates, the final amount was reduced to 12 papers that were explored for the production of this review.

Literature Review
In this section, the methods and image processing used in each researched work will be discussed. Two different forms of dust detection in solar panels are addressed: infrared cameras able to identify points of heat concentration in photovoltaic modules, and digital cameras that detect the state of the panel, if clean or dirty. Infrared images are able to identify hot spots on the surface of the panels. Phoolwani et al. (2020) analyzed at normal and some faulty condition the solar power meter using thermal camera to compare performances. Results were obtained at normal, partial shading and dust accumulated conditions using PV analyzer. Comparisons Development, v. 9, n. 8, e3219885107, 2020 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v9i8.5107 shape and it has software capable of detecting the state of the panel as clean or dirty, using the algorithm based on spectral decomposition developed in the previous article (Ramos et al., 2016), registering that panel.
The literature review allows to make comparisons about the methodologies. These comparisons are organized in Table 1. Some works do not specify certain details of the development, which may make it more difficult to reproduce. However, they explain parts of development of the method that can help and give inspiration for future research. The Table 1 is important, as it shows parameters to be considered when is required to start a work in this field of study. First, it is observed that new techniques continue to be performed every year. The cameras used can be either commercial or more specialized cameras, such as HD and those that capture infrared images. There are some ways to capture