With the increasing use of intelligent analysis technology, the existing intelligent analysis type of technology has not meet the gradually growing market demand, the security manufacturers should seek newer algorithms, richer business applications, overall system application, and strive to achieve in the application the breadth and depth of breakthrough. To this end, the security industry has begun to conduct research on a new generation of intelligent video analysis technology, new product form, the application model and system architecture, these new technologies, new products are gradually combined with the market, and explore new space for development. Case based intelligent monitoring requirements have been met, the new intelligent video technology focused toward the development of several directions: first, to enhance the visual sensory experience of technology, video preprocessing technologies including Image stabilization and image enhancement; second enhance the analysis of the accuracy of the technology, eyes recognition technology; Third, the application of technology to improve the system, including multi-ball machine interaction tracking; Fourth, for post-mortem analysis techniques, including image restoration, image summary retrieve and other technologies. Image jitter is often seen in the transport sector, the main reason is the higher frequency of elevated road monitoring surveillance cameras installation jitter slightly, and a significant jitter the car mobile monitoring due to the low frequency of the camera position changes . SD in analog era, mainly affecting the large magnification image screen in high-definition digital era, the screen will be significantly more than 20mm focal length jitter, which have a significant impact on the identification of monitoring content. Image stabilization is the main way to resolve several processing: software technology; complete acquisition sensor image, image processing buffer; reserved edge images to achieve image stabilization, digital zoom (image distortion or blurring of the central image) ; using reserved edge image, the image compensation, the stabilization effect (compensation area zoom, edge blur); enlarged to compensate for the digital image the image blur problem, use of a larger pixel of the image sensor; using direct physical pixel size the area of image dithering compensation, avoid image edge blur; the image judder technology is used in the front dome ip camera products, usually using a combination of hardware and software, one of the key technical points, to overcome the image edge corner of the imaginary The phenomenon of coke. Important issues facing the current monitoring system is affected by various confounders, so that the system can not be all-weather, all-day nowadays run, the reliability of the system is low, LCC value is not economic. These confounding factors including meteorological conditions, such as daytime virtual environment light, flood light, backlight, low-light environments, monitoring the interference of strong backlight scene, fog, haze and other atmospheric scattering environment interference, rain, snow, dust, etc. the impact. Image enhancement technology to solve a problem, visual video source through the algorithm to improve the handling, improve quality, and to improve the clarity of the image to improve the visual distinguished the original low-quality images to meet the monitoring needs to reach a clear distinguish the degree. Technical Realization of image enhancement, projecting the original image of the contents of the key observations, inhibiting the contents of the non-key observation, through the pixel gray value calculation processing to generate a new image, and to improve the visual effect. Key technical issues of image enhancement: different times for different adaptive, varying degrees of interference factors; effective resolution screen fully adaptive to different depth of field, angle of view, target content; retain local details; the screen contrast soft, balanced color transition moderate; bright and dark contrast color constancy, not scene changes affect; under normal circumstances does not affect the picture quality; the unreinforced screen without distortion; product form, covering the end of the show from the front end coding background image enhancement processing. Product positioning, can be divided into two directions: the universal product focus did not fall under the normal climate qualitative improvement in harsh climates, but not the best, can only be integrated into various devices as standard; The focus of professional-grade products to achieve the best results to achieve in a bad situation, as special equipment, such as fog dedicated, deep-sea dedicated low-light special, military dedicated application integration in special industries. Binocular stereo vision technology's core purpose is to improve the recognition accuracy. As with the three-dimensional geometry of the object information in the field of view formed by the stereo vision techniques, it is possible to effectively set the detection rule, to exclude light, shadow and other interfering factors, a substantial increase in the accuracy intelligent analysis. If HD technology to enhance the available pixels to improve the accuracy of the analysis is a tactical move, binocular stereo vision technology for video analysis accuracy is strategic. Binocular stereo vision technology is based on the principle of parallax, and use of the imaging apparatus to obtain two images of the measured object, by calculating the deviation of the position of the image between corresponding points, to obtain a method of the object three-dimensional geometrical information from different locations. Dual-camera or multi-camera, high-precision three-dimensional position of the free movement of the body of the field of view of space coordinates and attitude measurement, determine the centroid of the moving target position, and high-precision tracking of moving targets based on the results of calibration. As is able to recognize the three-dimensional coordinates of the target, attitude, the relative distance, and the background environment space distance, it can adapt to the complexity of tracking the background environment.
|