|
Coherent laser radar can simultaneously into an intensity image and range image, and the intensity image and range image information fusion, it is possible to form a complete three-dimensional information, which makes coherent laser radar target identification and targeting, and so has a great prospect. At home and abroad for LIDAR intensity of research has been done like more work, but the distance is not like a lot of research, this article will focus on the distance of image processing research. This paper studied the mechanism of coherent laser radar imaging, imaging mechanism through meticulous understanding of the research results at home and abroad, the noise from the image analysis of the reasons for the formation and the establishment of its noise model. Based on the noise model, have been collected from the simulation like to add noise to the following work provides an important basis of experimental data. Secondly, the distance like a lost mainly by the impact of information and outliers. Lost information exists, resulting from complicated as background; while outliers, led to errors in the distance like the depth of the information that emerged from unusual. This data collection to establish the experimental scene is an ideal situation, so add in the image distance simulation noise of only abnormal noise, so this paper is made from unusual noise suppression studies. The paper first summarizes the commonly used method of noise suppression algorithms and features, combined with the data from the image characteristics, proposed a general principle, namely that the normal pixel is always surrounded by abnormal pixels, this article will be referred to as bracketing guidelines. This criterion, combined with median filtering and weighted average filter, containing more distance from abnormal noise as conducted noise rejection. This method is not only effective for noise suppression, but also better preserve the image details, in the noise suppression and detail preservation good compromise between. But because of its complexity of the algorithm, but also makes the algorithm more time-consuming, so this algorithm timeliness there is still room for improvement. Again, this article also extracted from the image edges were studied. This paper compares the classical edge detection algorithm, pointing out that the commonly used classical algorithm for edge detection have a inaccuracies and incompleteness in fact, with little significance. This article will image segmentation and edge detection combined theory was proposed based on image segmentation and Canny operator edge extraction algorithm through segmentation theory classic edge detection algorithm to make up for the lack of additional information is defined as the goal of the edge contour information, The detailed information in the target filter. Through simulation experiments, we can see that the detection effect is very good, relatively accurate contour information, closure, suitable for image segmentation.
|