Visão computacional para estimativa de comportamento de aglomeração de galinhas poedeiras
Arquivos
Data
2019-02-22
Autores
Lopes, Francisco Augusto Alves
Título da Revista
ISSN da Revista
Título de Volume
Editor
Universidade Estadual Paulista (Unesp)
Resumo
As aves são animais homeotérmicos e manter a temperatura corporal dentro de limites da termoneutralidade proporcionam maior produtividade e melhor bem-estar. Quando a temperatura do ambiente está fora da zona de termoneutralidade, as aves expressam comportamentos individuais e de grupo que acusam esse desconforto. A hipótese deste trabalho é que o comportamento de aglomeração das aves possa ser utilizado para expressar seu nível de conforto térmico. O objetivo deste trabalho foi desenvolver um algoritmo computacional para processamento e análise de imagens de poedeiras criadas sobre cama de maravalha capaz de identificar o grau de aglomeração das aves em diferentes condições de conforto térmico. Foram objetivos específicos propor um índice de aglomeração baseado em medidas obtidas na análise das imagens e confirmar as diferenças deste índice de aglomeração em diferentes condições de temperatura do ambiente de criação. O algoritmo foi desenvolvido em ambiente MatLab® e foi dividido em duas etapas: processamento e análise da imagem. Foi realizado um experimento em três aviários de escala reduzida e distorcida com grupos de 10 aves em cada aviário. As aves foram monitoradas por câmeras de vídeo instaladas no teto dos aviários. Os vídeos foram processados frame a frame pelo algoritmo e medidas de área, perímetro e distâncias entre os centros de massa dos aglomerados de aves foram obtidas na análise das imagens. A partir dessas medidas foi proposto um índice de aglomeração que foi usado para calcular o nível de aglomeração de cada frame de dois vídeos obtidos em condições térmicas distintas: 21ºC e 32ºC. Os resultados obtidos de índice de aglomeração nas duas condições de temperatura foram submetidos ao teste T-student que confirmou diferença nas duas situações, validando dessa forma o índice de aglomeração proposto. Foi feita a repetição do experimento em condições de ambiente térmico diferentes e constatou-se que o algoritmo desenvolvido não respondeu adequadamente à troca da cama e às variações de iluminação que se apresentaram, o que demonstra que mais ajustes deverão ser realizados. Conclui-se que o algoritmo desenvolvido foi capaz de identificar as aves do restante da imagem por meio do processamento digital das imagens, o Índice de Aglomeração. pode ser aplicado em sistemas de visão computacional para estimar o conforto térmico de um grupo de galinhas poedeiras e que técnicas de visão computacional para a análise de comportamento de grupo são promissores para monitorar e indicar o bem-estar térmico de galinhas poedeiras.
Birds are homeothermic animals and maintaining body temperature within the limits of thermoneutrality provide greater productivity and better well-being. When the ambient temperature is outside the thermoneutral zone, the birds express individual and group behaviors that accuse this discomfort. The hypothesis of this work is that the agglomeration behavior of the birds can be used to express their level of thermal comfort. The objective of this work was to develop a computational algorithm for image processing and analysis of laying hens on a bed of marbles capable of identifying the degree of agglomeration of the birds in different conditions of thermal comfort. Specific objectives were to propose an agglomeration index based on measurements obtained in the analysis of the images and to confirm the differences of this agglomeration index in different temperature conditions of the breeding environment. The algorithm was developed in MatLab® environment and was divided into two stages: image processing and analysis. An experiment was carried out on three small scale and distorted aviaries with groups of 10 birds in each aviary. The birds were monitored by video cameras installed on the ceiling of the aviaries. The videos were processed frame by frame by the algorithm and measurements of area, perimeter and distances between the centers of mass of birds were obtained in the analysis of the images. From these measurements an agglomeration index was proposed that was used to calculate the level of agglomeration of each frame of two videos obtained under different thermal conditions: 21ºC and 32ºC. The results of agglomeration index in the two temperature conditions were submitted to the T-student test, which confirmed the difference in the two situations, thus validating the proposed agglomeration index. The experiment was repeated in different thermal environment conditions and it was found that the developed algorithm did not respond adequately to the bed change and the lighting variations that were presented, which shows that more adjustments should be made. It is concluded that the developed algorithm was able to identify the birds of the rest of the image by means of the digital image processing, the Agglomeration Index. can be applied in computer vision systems to estimate the thermal comfort of a group of laying hens and that computer vision techniques for group behavior analysis are promising to monitor and indicate the thermal well-being of laying hens.
Birds are homeothermic animals and maintaining body temperature within the limits of thermoneutrality provide greater productivity and better well-being. When the ambient temperature is outside the thermoneutral zone, the birds express individual and group behaviors that accuse this discomfort. The hypothesis of this work is that the agglomeration behavior of the birds can be used to express their level of thermal comfort. The objective of this work was to develop a computational algorithm for image processing and analysis of laying hens on a bed of marbles capable of identifying the degree of agglomeration of the birds in different conditions of thermal comfort. Specific objectives were to propose an agglomeration index based on measurements obtained in the analysis of the images and to confirm the differences of this agglomeration index in different temperature conditions of the breeding environment. The algorithm was developed in MatLab® environment and was divided into two stages: image processing and analysis. An experiment was carried out on three small scale and distorted aviaries with groups of 10 birds in each aviary. The birds were monitored by video cameras installed on the ceiling of the aviaries. The videos were processed frame by frame by the algorithm and measurements of area, perimeter and distances between the centers of mass of birds were obtained in the analysis of the images. From these measurements an agglomeration index was proposed that was used to calculate the level of agglomeration of each frame of two videos obtained under different thermal conditions: 21ºC and 32ºC. The results of agglomeration index in the two temperature conditions were submitted to the T-student test, which confirmed the difference in the two situations, thus validating the proposed agglomeration index. The experiment was repeated in different thermal environment conditions and it was found that the developed algorithm did not respond adequately to the bed change and the lighting variations that were presented, which shows that more adjustments should be made. It is concluded that the developed algorithm was able to identify the birds of the rest of the image by means of the digital image processing, the Agglomeration Index. can be applied in computer vision systems to estimate the thermal comfort of a group of laying hens and that computer vision techniques for group behavior analysis are promising to monitor and indicate the thermal well-being of laying hens.
Descrição
Palavras-chave
análise de imagens, bem-estar animal, conforto térmico, zootecnia de precisão.