Research On Target Tracking Technology In Sea Cucumber Automatic Fishing

With the improvement of living standards,people on sea cucumber and other seafood demand is growing.Sea cucumber farming depth is generally 10?40m,and mainly collected in the way of artificial diving operations for finished fishing.But diving-type fishing is not only low yield,high cost,and will seriously damage the health of divers.In this paper,underwater robot is used to study the technology of machine vision tracking and location for sea cucumber fishing,the main research includes the following,During the pre-research period,to study the video in the laboratory environment simulation of the seabed environment as the research object,an improved Mean-Shift tracking algorithm is proposed.The main idea is to extract the contours of sea cucumbers by using the Mean-Shift algorithm to iteratively search the contour centroids to achieve position tracking and scale calculation.After obtaining the real underwater sea cucumber video,we first use the Mean-Shift algorithm to track the sea cucumber.The actual underwater image contrast is too low and has a poor quality to be accurate to get the profile of sea cucumber.For this reason,the de-fog algorithm for image processing is used first,and then use the Mean-Shift algorithm to achieve the sea cucumber tracking.Because the Defogging Mean-Shift algorithm has a low tracking efficiency.In order to improve the tracking efficiency,an underwater sea cucumber tracking algorithm based on kernel correlation filtering(KCF)is proposed.The main idea of the algorithm is to track the two heads of sea cucumber,which is not only improve the tracking efficiency,but also improve the accuracy of position and scale.By using the improved kernelized correlation filters(KCF)tracking algorithm,the image processing speed can reach 42fps,and the average tracking accuracy can be achieved to 88.98%,which has a highly success rate.Therefore,all of the three algorithms proposed in this paper can meet the accuracy requirements of tracking,and these algorithms can achieved the target sea cucumber in complex environment.Among these algorithms,KCF tracking algorithm has the highest speed for tracking target sea cucumber identification and positioning,meanwhile it not only can meet the fast feedback requirements of control system to find the target sea cucumber,but also have a good stability,a high success rate and a good robustness.