Study On Detection Method Of Dead Chicken In Unmanned Chicken Farm

The rapid development of information technology,the gradual maturity of artificial intelligence technology,and the improvement of image processing technology have accelerated the pace of people entering the intelligent production lifestyle.As an important part of human production methods,poultry breeding industry has become an important topic for many experts and scholars to realize the intelligence of breeding industry.Based on this situation,this paper proposes a method for detecting dead chickens in unmanned chicken farms.Image processing technology is used to process and identify the images collected by the breeding robots,to determine whether there are dead chickens in the pictures,to classify the chicken images,and to help the farmers quickly find dead chickens in the chicken cages.This technology not only helps the farmers to improve work efficiency,reduce unnecessary manpower,but also avoids unnecessary pathogen infection caused by excessive contact between chickens in the chicken farm and the outside world.At the same time,it improves the working environment of the farmers and raises personnel.It is not necessary to stay in a chicken farm with poor air quality for a long time to find dead chickens.The detection method of dead chicken in unmanned chicken farm proposed in this paper is divided into two parts,namely image processing part and image recognition for classification.The image processing section processes the chicken image collected by the breeding robot.It mainly includes the enhancement algorithm of chicken image,denoising algorithm,segmentation algorithm and edge detection algorithm.Aiming at the characteristics of chicken images taken in unmanned chicken farms,this paper proposes an image enhancement algorithm combining Gamma correction and adaptive histogram equalization to enhance chicken images;using wavelet separation and bilateral filtering combined The noise method is used to denoise the chicken image.The chicken image is segmented based on the threshold segmentation algorithm.An edge detection algorithm based on morphological gradient is proposed to detect the edge of the chicken image.The image recognition classification section is based on the color feature and the shape feature extraction method to obtain the chicken claw shape feature and the chicken crown color feature to classify the chicken image.Firstly,according to the shape feature and color feature of the processed image,the chicken image training data set is classified and trained,and the trained model can identify the state(health or death)of the chicken in the image and classify it;secondly,utilize The chicken image test data set tests the trained model.In the classification processing of chicken images,this paper proposes a method based on Lib SVM for detecting dead chickens.By observing the collected images,only the cockscomb is red in the image,and when the cockscomb is close to the ground,it can be considered that the chicken is dead at this time;the angle of the image collection can be seen in the chicken feet of most chickens in the chicken cage,according to the live The morphological characteristics of chicken feet and dead chicken feet can identify the state of the chicken and be used for the classification of chicken images.Lib SVM is an open source library for support vector machines,mainly used for classification and regression.Lib SVM has the characteristics of simple operation,easy to use,fast and effective,and relatively few adjustments to the parameters involved in SVM.It is suitable for small sample classification processing.This paper classifies chicken images based on the shape and color characteristics of Lib SVM and chicken images.In this paper,the image processing algorithm is compared and compared with the processing results of the classical image processing algorithm.The image histogram and signal-to-noise ratio are used to show that the chicken image processing method proposed in this paper is suitable for breeding robots.Image.Experiments were carried out on the detection method of dead chicken based on Lib SVM.Through the recording and sorting of experimental results,the classification accuracy of chicken images based on Lib SVM classification method can reach more than 90%.