Research Of Some Key Algorithms In Micro-Displacement Dynamic Image Analysis System

The oil is a kind of important and gradually dried up energy. The water/oil displacement experiment based on rock pore micro-model is a kind of important means to study the oil's filtration characteristic and the distributing rule of remaining oil. Combining dynamic image analyzing technique with micro-model displacement technique, we have developed the microscopic displacement dynamic image analysis system successfully. This system can realize many functions such as recording dynamic images real-timely and analyzing those images quantitatively:Record the change of the displaced target's characteristics real-timely and all kinds of parameters of the serial images.Measure, compute the basic characteristic parameters of the serial images recorded.Diagrammatize the dynamic variety regulation of the characteristic parameters, and calculate their mathematics expressions consisting of time and coordinates.Detect, track the moving oil-beads in the some rock pore, and sketch their track-line in the macroscopic and microscopic images contrastively.In this paper, we just make researches on some key techniques of this microscopic displacement image analysis system by analyzing dynamic images and doing many experiments carefully, present and realize the following algorithms or methods.1. This thesis presents a novel method of image background generation based on the stat-average and blind correction. This method can't result in object's movingshadow which appears in the method of N fames average and also do not need the artificial setting or adjustment every kind of parameters or factors in image background generation. The experiments indicate that this method have many advantages such as the lower computational complexity and no artificial interference etc.2. Feature selection is the key part in image matching based on features. A new image registration algorithms based on the rock grain characteristic is proposed. It has a lower computational complexity and has no limit to the ratio of the high-low image resolution theoretically, as long as the rock grain characteristic can be obtained in the low-resolution macroscopic image well. Its efficiency is validated by the experiments.3. Had discussed the moving character of the particular oil-bead and the existing algorithms about the detection and tracking technique of motion object, we improve the oil-bead detection and picking-up method, Put forward a novel method of the oil-bead tracking based on its invariable characteristic. The experiments indicates, the mentioned methods can reduce the search space, accelerate the characteristic matching and tracking , have good adaptability to the microscopic serial images in the developed system.