An intelligent parking assist device based on omnibearing computer vision, including visual sensor for parking assist decision microprocessor and a display unit for displaying the outside of the vehicle, the visual sensor is connected with the microprocessor, the microprocessor includes: the image display unit, signal acquisition for vision sensor, and output to the display unit; vision sensor is characterized in that: the top of the omnidirectional vision sensor, omnidirectional vision sensors installed in the vehicles in the middle, the full range of visual sensor includes a convex catadioptric mirror around the object in the field of vehicle reflection, convex catadioptric mirror face down, to prevent the light refraction and the light saturated black cone, a black cone fixed on the protruding part of the folding mirror in The heart, for the transparent cylinder supporting convex catadioptric mirror and convex mirror imaging shooting body camera for cameras, the convex mirror at the top; the microprocessor also includes: sensor calibration module, used for correspondence like and video image obtained by the road to build the space image map; the image module for omnidirectional vision sensor acquisition of video image, the circular read for panoramic histogram; obstacles around the vehicle edge detection module, used by the Sobel operator for edge detection, using the set size of the template, calculate the gradient in each template in the image, select the calculation value is greater than the gradient gradient the amplitude of the points as edge points, each edge point connected into edge Edge position; vehicle obstacles around the nearest edge distance calculation module, used in the right and left direction, with the length of the vehicle as test line, alarm distance from the vehicle width plus as the start of the test line, respectively, from about outward direction detection, when the detection line touched the edge line detection line and obstacle preset the intersection of pixels when judging the obstacles and vehicle distance on the right and left direction; in the direction, with the width of the vehicle as test line, alarm distance from the length of the vehicle and began as a detection line, respectively from the outward direction before and after the test, when the test line touched the edge line line detection and obstacle preset pixel number of the intersection, the judge is the obstacle and the distance to the nearest vehicle in the longitudinal direction; The obstacle alarm module is used to judge whether the lower limit of the nearest distance is less than a preset value, such as less than the limit value, the alarm instruction to the alarm; reversing the knowledge base, used to store the driver reversing skills, skills, as well as between the various current position and the target position of the backing track line, call for the generation of backing track line; parking specify the location for the user through the user interface module, input hope parking position, using the image information around the omni-directional vision sensor of the user in the image information display on the target position on the touch screen; generating reversing trajectory module for vehicle position through the omni-directional vision sensor can detect the current user. By entering the target position of the user interface, the automatic generation of an inverted The car tracks, backing the envelope according to the corresponding reverse trajectory, reversing the envelope is greater than the vehicle can travel path curvature minimum, and determine the image edge reversing envelope and obstacles such as whether the intersection, the intersection will not determine the reverse track trace; otherwise, the intersection modified reverse trajectory; steering wheel angle action calculation module, according to the reversing trajectory line generated by the target path, the key points in the backing track on the line of driving down angle P, the relationship between target path angle: P and steering angle decreases with delta H the formula (22), P = psi: Tan = \1\ (Y: P / X: P) = Y: P / X: P: K = s * 8: h * t: P / 2 (22) in the formula, K: S = G = gamma, Delta, down:: G: * s is a constant, estimated to reach a critical point in the reversing speed situation when the time: t p; and the steering wheel angle Delta: H output to the display unit; reversing trajectory correction module, used to recalculate the backing track line deviation in the vehicle deviates from the backing track line, detection of the key points in reversing the track line whether the deviation, if the detected offset value exceeds the preset range according to the current position of the vehicle, to re generate the backing track line.
【技术实现步骤摘要】
本专利技术涉及全方位视觉传感器技术、图像识别技术、计算机控制技术在车辆停车方面的应用,尤其是一种智能这车辅助装置。
技术介绍
随着我国汽车产业的发展和人们生活水平的不断提高,一方面汽车的数量逐年增加,每年以700万台速度增长,使得我国的公路、街道及停车场拥挤不堪,可转动的空间越来越少;另一方面,车辆绝对数的增长造成了新司机及非专职司机越来越多,因倒车引起的纠纷越来越多。在2002年汽车事故的发生比例中,倒车引起的事故占21%,倒车成为令人们头痛的一项任务,即使是经验丰富的司机也在抱怨倒车是件费力费神的事。改善倒车遇到的窘境被越来越多的人所关注,人们对汽车操纵的便捷性提出了更高的要求,希望有种装置能够解决汽车倒车给人们带来的不便,消除驾驶中的不安全因素,可将车快速准确地停放到指定的位置。汽车倒车中存在的困难,其一是来自于驾驶者坐在驾驶座位上无法完全了解汽车四周特别是后方的环境。驾驶者只能依赖后视镜观察后方障碍物,后视镜受其位置的限制,视野狭窄、清晰度差,根本无法达到倒车中后视的目的。其二倒车时驾驶者同时要兼顾车辆前方、两侧的情况,必须扭身回头观察车辆后面的情况,体力和脑力消耗过大,易产生不安全因素。其三倒车是一个复杂的工程,它依赖于驾驶者的驾驶经验、驾驶技巧及反应灵敏程度,任一环节出问题都导致驾驶员无法快速准确地完成倒车任务。中国专利技术专利00100982.6公开了一种具有多路显示功能的倒车雷达装置。该方案采用在车尾适当位置分设超声波感应器及摄像机,二者分别连接至一倒车雷达主机及图像处理模组,倒车雷达主机亦与图像处理模组连线,图像处理模组输出端则与位于驾 ...
【技术保护点】
【技术特征摘要】
【专利技术属性】
技术研发人员:汤一平,严海东,柳圣军,贺武杰,金海明,周思宗,
申请(专利权)人:浙江工业大学,
类型:发明
国别省市:
还没有人留言评论。发表了对其他浏览者有用的留言会获得科技券。