Moving Object Detection And Semi-singularity Avoidance For The 3-axis Tracking Pedestal

Moving target detection and targeting technology are widely used.A typical application is the photoelectric tracking and aiming platform.This platform is the core component of many sophisticated communication and detection devices.A high quality photoelectric tracking and aiming platform must be able to accurately identify the target from the complex background.In addition,the control system for the aiming mechanism must have good rapidity,accuracy and stability.Therefore,how to perform target detection and tracking in a highly complex environment are the key problems in the design of a photoelectric detection and targeting system.As an important aspect of photoelectric detection technology,moving target detection based on a CCD sensor can provide object information such as shape,color,attitude,action and behavior,which play an important role in an intelligent detection system.However,due to the complexity of the natural environment,moving object detection based on a CCD sensor has many challenges,such as the existence of background illumination changes and dynamic environment such as glittering water or swaying trees.The existence of these difficulties has a negative impact on the detection results,and makes it much more difficult to track and analyze the objects.Therefore,how to construct a robust and accurate moving object detection method have been widely concerned for a long time.For a photoelectric tracking and aiming platform,the aiming mechanism with two rotational joints is widely used.However,this kind of mechanism is suffered from the keyhole problem.Therefore,people try to add an additional rotating joint to the aiming system,making it a three axis pointing mechanism.It is because of this extra degree of freedom of the system that the system can avoided the key-hole problem.However,we have found that this kind of multi-axis mechanism may exist semi singularity owing to rotational limits.Unfortunately,existing control strategies for avoiding the rotation limit and common singularity are not effective in the avoidance of semi singularity.Therefore,how to design an effective online semi singularity avoidance strategy for the aiming mechanism becomes a critical issue in the control of a pointing mechanism.In order to deal with the above technical difficulties,many attempts are taken in moving target detection under dynamic environment and the semi singularity avoidance strategy of the three-axis pointing system.The main contributions of work are given as follows:(1)we have found that the color model of a classical background model is only suitable for the single light source disturbance.To extend this model to more general cases,we propose an arbitrary cylinder color model with a highly efficient updating strategy.The classical cylindrical color model can be seen as a special case of the proposed model.Experimental results show that,with no loss of real-time performance,the proposed model reduces the wrong classification rate of the cylinder color model by more than fifty percent.(2)Since a conventional histogram is too sensitive to noise in modeling a background pixel,we propose a novel pixel-modeling approach using histograms based on strong uniform fuzzy partitions.In the proposed method,the temporal distribution of pixel values is represented by a histogram based on a triangular partition.The threshold for background segmentation is set adaptively according to the shape of the histogram.Histogram accumulation is controlled adaptively by a fuzzy controller under a supervised learning framework.Benefiting from the adaptive scheme,without parameter tuning,the proposed algorithm functions well across a wide spectrum of challenging environments.Further,it outperforms other state-of-the-art methods by a significant margin.(3)We propose a novel adaptive maintenance scheme for the codebook-based background subtraction algorithm.With this technique,the accuracy and efficiency of the model are significantly improved.In the proposed method,we develop an equalqualification updating strategy to replace the maximum-negative-run-length-based filtering strategy.Further,we substitute the cache-based foreground learning process with a random updating scheme.These modifications not only preserve the accuracy of the codebook model but also significantly reduce the number of parameters used in the maintenance scheme.In the modified framework,parameters that are scenario-sensitive are identified through extensive experiments and analysis.Furthermore,adaptive methods are proposed for them.The proposed method ensures the best performance of the system across a variety of complex scenarios.In our experiments,it is demonstrated that,despite using only color information,the proposed method outperforms the majority of the solutions by a significant margin.(4)We have found that,owing to their rotary limits,three-axis pedestals have semi singular postures.In these postures,they cannot follow targets moving in certain directions.More critically,when using common unilateral constraint avoidance approaches,there always exists a target movement for which the pedestal will be guided into semi singular configurations.This is a fatal defect when tracking non-cooperative targets.To solve this serious problem,we propose a novel online solution for semi singularity avoidance.A reduced configuration space is developed so that the pedestal does not have semi singularity.We propose a general method to derive such a configuration space for the three-axis tracking pedestal under arbitrary elevation and heeling-motion limits.Experimental results clearly show that,with the proposed approach,the pedestal can effectively avoid semi singularity with a very smooth posture trajectory.