Research And Realization On Segmentation Algorithm For Rock Particles Image

Rock particles image segmentation is the important pre-work of core analysis. The result of the core analysis can provide data for core analysis, and core analysis can be the data source for petroleum exploration. Petroleum plays an important role in economic development and national defense. And the demand for oil is on the increase. On the other hand, the petroleum is one-off, non-renewable resource. So modern petroleum analysis system is necessary for petroleum exploitation. Computer software can be used to retrieve core data of rock particles image, we can convert manual input of the data to automatic analysis extraction by powerful computing capability of the computer, so the aid of computer software can be used to extract intuitively core data of rock particles image. Its results can be used for core analysis. Rational division of rock particles and accurate extraction of core particles have important practical significance.This paper is in order to solve the problem of image segmentation of the core particle in core analysis. The image is understood as a digital code containing useful information by computer. Image processing is the use of computer for effective processing of these digital codes to meet the application requirements,While image segmentation is to divide the image into several disjoint regions by the feature containing in the processing images such as intensity, color, texture, shape and so on, to distribute those with the similar or consistent characteristics into a same region, but show significant differences in different regions. Simply speaking, it refers to extract the object in the image from the background so as to process the target. People have always attached great importance to the research on the field of image segmentation for many years, and new algorithms as well as the improvement of the image segmentation algorithms have been put forward. Due to the fact that image segmentation effect depends mainly on the specific image features such as image, texture, gray, noise and so on, therefore, there is not a single universal image segmentation algorithm which can be applied to effective segmentation of all images. Currently, image segmentation methods which are widely used at home and abroad generally include threshold segmentation, segmentation based on edge, segmentation based on deformable model, segmentation based on region growing, segmentation based on genetic algorithm, segmentation based on artificial neural networks. Based on the mainstream algorithm of image segmentation, combining with the characteristics of rock particle image itself, this paper puts forward a new method to segment rock particles image and it has turned out to achieve a good segmentation effect expected. The main content of this paper is completed as follows:1. Study on the rock particle segmentation based on segmentation threshold algorithm. Focus on the algorithm principle, procedures and experimental results of global threshold method, local threshold method, and adaptive threshold method. It can be seen from the experiment process and the effect that the theory of rock particle segmentation algorithm based on threshold segmentation is brief and concise when threshold method is applied to the rock particle image segmentation. But the segmentation effect is not obvious of the target and image with no obvious background contrast, which s resulted from the fact that the threshold algorithm basically only use the pixel gray value and not make full use of the space information of the image. Finally, segmentation algorithm based on region growing, split and merge is introduces. Segmentation algorithm based on region make full use of the pixel value and space information but is limited by seed, growth standards or splitting and merging criterion limit, so some other image processing algorithms are often needed to be combined to achieve the segmentation effect. At the same time the introduction of segmentation algorithm based on region provides theoretical basis for merging region of a new algorithm. It can be seen obviously from the study on the threshold and regional segmentation algorithm that pixel value and space information of the image information should be made full use to solve the segmentation problem of the rock particle image.2. Study of detection technology based on edge of the serial and parallel on the rock particle image segmentation. Detection technology based on edge of the serial and parallel is to extract rock particle, and closed and continuous edge can also be obtained even in case of huge image noise. Differential operator method is widely used in the parallel detection technology, in which first-order differential operator and the second-order differential operator is more frequently used. Oberts, Sobel, Prewitt, Kirsh, Canny and some other operators are commonly used in the first-order differential operator, while Laplace, Log and some other operators are commonly used in the second-order differential operator. Differential operator is widely applied not only to the edge extraction, but also to pre-extraction process of image gradient in the preprocess of image processing.3. Study on combination of specific segmentation theory for rock particle segmentation algorithm. FCM clustering segmentation algorithm based on clustering analysis algorithm has been successfully applied in the field of image segmentation. Processing technology of digital morphology and watershed segmentation algorithm are mainly focused on. There are some basic operations such as corrosion, expansion, open and close in mathematical morphology. Based on these operations morphological thinning and morphological filtering technique are achieved. Watershed segmentation algorithm is derived from the morphology, with the advantage of using the spatial information and fast processing speed, but there is a serious problem of over segmentation remaining to be solved.4. Put forward a new method of rock particle segmentation. The new method of rock particle segmentation is based on the analysis and summary of image segmentation based on threshold, region, edge and specific theory. The new method mainly regards digital morphology and watershed algorithm as the theoretical basis, learning from the ideas of edge segmentation for keeping the main information, combined with the post-processing of regional division theory. The new method uses the morphological opening and closing reconstruction filter for de-noising the original rock particle image to achieve a good de-noising effect. Rock particle and image background are respectively marked out by background marker and foreground marker, which provides referential standard for watershed segmentation. Avoiding the repeated access to the operation of the queue, the segmentation by watershed algorithm based on intuitive watershed has a certain improved efficiency in the algorithm. Finally, region merging is processed to the segmented image. It can be seen from effect of final rock particle image segmentation that the new method effectively solves the over-segmentation problem of traditional watershed algorithm, and can extract the rock particle accurately, achieving the expected processed effect. The final segmentation effect can meet the demand for rock analysis.