会员体验
专利管家(专利管理)
工作空间(专利管理)
风险监控(情报监控)
数据分析(专利分析)
侵权分析(诉讼无效)
联系我们
交流群
官方交流:
QQ群: 891211   
微信请扫码    >>>
现在联系顾问~
热词
    • 5. 发明公开
    • 표본 프로파일 정보 기반 움직임 추정장치 및 방법
    • 基于抽样概况的运动估计的装置和方法
    • KR1020130011608A
    • 2013-01-30
    • KR1020110072890
    • 2011-07-22
    • 에스케이플래닛 주식회사
    • 배태면
    • H04N19/51H04N19/14
    • H04N19/51G06T7/13G06T7/181G06T7/223H04N19/14H04N19/176
    • PURPOSE: A motion estimating device based on sample profile information and a method thereof are provided to reduce calculation for motion estimation by detecting an edge of a reference image and a current image, comparing matching errors of a vertical line and a horizontal line including the most edge information, and estimating a motion vector. CONSTITUTION: An edge image generating unit(110) detects an edge corresponding to an outline of an object in a current image. An image analyzing unit(120) receives a generated edge image. The image analyzing unit detects a horizontal max line and a vertical max line including the most edge information in a motion estimation target block. A motion vector estimating unit(130) calculates a block matching error between the current image and a reference image based on the horizontal max line and the vertical max line. The motion vector estimating unit estimates a coordinate difference of a block having the smallest difference as motion vector. [Reference numerals] (110) Edge image generating unit; (120) Image analyzing unit; (130) Motion vector estimating unit; (140) Motion vector storage unit; (AA) Current image; (BB) Reference image
    • 目的:提供一种基于样本分布信息的运动估计装置及其方法,通过检测参考图像和当前图像的边缘来减少运动估计的计算,比较垂直线和包括最多的垂直线的水平线的匹配误差 边缘信息,以及估计运动矢量。 构成:边缘图像生成单元(110)检测与当前图像中的对象的轮廓对应的边缘。 图像分析单元(120)接收生成的边缘图像。 图像分析单元检测包括运动估计目标块中的最边缘信息的水平最大线和垂直最大线。 运动矢量估计单元(130)基于水平最大线和垂直最大线来计算当前图像和参考图像之间的块匹配误差。 运动矢量估计单元将具有最小差的块的坐标差估计为运动矢量。 (附图标记)(110)边缘图像生成单元; (120)图像分析单元; (130)运动矢量估计单元; (140)运动矢量存储单元; (AA)当前图像; (BB)参考图像
    • 6. 发明公开
    • 이미지 처리 장치, 이미지 처리 장치의 에지 향상 방법 및 에지 향상 방법이 기록된 기록 매체
    • 用于处理图像信号的装置,用于增强图像信号处理装置边缘的方法和用于提高边缘的完美方法的记录介质
    • KR1020100079324A
    • 2010-07-08
    • KR1020080137763
    • 2008-12-31
    • 엠텍비젼 주식회사
    • 이호영
    • H04N5/208
    • H04N5/208G06T5/20G06T7/181G06T2207/10024G06T2207/20016G06T2207/20192H04N5/2352
    • PURPOSE: A device for processing an image signal, method for enhancing edge of an image signal processing device and a recorded medium for recording the same are provided to effectively perform an edge improvement of an image. CONSTITUTION: If a second filtering value is smaller than a region boundary value, a filtering strength applying unit(240) determines surface improvement correction value corresponding to low frequency intensity value between inputted image luminance value and a first filtering value based on the filtering strength applying unit. If the second filtering value is bigger than the region boundary value, the filtering intensity applying unit obtains an edge table value corresponding to the second filtering value in an edge table. The filtering intensity applying unit determines an edge improvement correction value based on the obtained edge table value and the edge improvement correction intensity.
    • 目的:提供一种用于处理图像信号的装置,用于增强图像信号处理装置的边缘的方法和用于记录图像信号处理装置的记录介质,以有效地执行图像的边缘改进。 构成:如果第二滤波值小于区域边界值,则滤波强度施加单元(240)基于施加的滤波强度来确定与输入的图像亮度值和第一滤波值之间的低频强度值对应的表面改善校正值 单元。 如果第二滤波值大于区域边界值,则滤波强度应用单元获得与边缘表中的第二滤波值对应的边缘表值。 滤波强度施加单元基于获得的边缘表值和边缘改善校正强度来确定边缘改善校正值。
    • 7. 发明公开
    • 윤곽을 따르는 트래커를 자동적으로 결정하기 위한 방법및 그 방법을 실시하는 프로그램을 기억한 기억 매체
    • 用于自动决定跟踪器的方法,以及用于执行方法的存储介质存储程序
    • KR1020010014871A
    • 2001-02-26
    • KR1020000024028
    • 2000-05-04
    • 인터내셔널 비지네스 머신즈 코포레이션
    • 호리우치요시오류토마사유키
    • G06F9/44
    • G06T7/12G06T7/181G06T2207/20104G06T2207/20168
    • PURPOSE: To easily detect the outline of a desired area to be extracted from an image without tracing the outline by deciding a combination of tracker candidate points to minimize the cost by means of a dynamic programming method with the selected one of plural initial points defined as a start point. CONSTITUTION: The initial points are positioned on a boundary of an initial area that is set to include a desired object to be extracted from an image (200). The distance between two adjacent initial points is set at the value less than the threshold by increasing the new initial points on the boundary (201). The inside of the initial area is scanned at every initial point to decide the tracker candidates (202). A combination of tracker candidates is decided to minimize the cost by means of a dynamic programming method among those combinations of tracker candidates and also to best show the outline of the desired object to be extracted (203). Then the tracker candidates are thinned to decide a tracker so that the distance set between the trackers is approximately equal to DT (204).
    • 目的:通过决定一个跟踪者候选点的组合来轻松地检测要从图像中提取的所需区域的轮廓,以通过动态规划方法来最小化成本,将多个初始点中的一个定义为 一个起点 构成:初始点位于被设置为包括要从图像提取的期望对象的初始区域的边界上(200)。 通过增加边界上的新的初始点(201)将两个相邻初始点之间的距离设置为小于阈值的值。 在每个初始点扫描初始区域的内部以决定跟踪者候选(202)。 确定跟踪者候选者的组合,通过在所述跟踪者候选者的组合中的动态规划方法来最小化成本,并且最好地示出要提取的期望对象的轮廓(203)。 然后,将跟踪者候选者减薄以确定跟踪器,使得在跟踪器之间设置的距离近似等于DT(204)。
    • 8. 发明授权
    • 차선 인식 장치 및 그 방법
    • 用于检测LANE的装置及其方法
    • KR101578434B1
    • 2015-12-17
    • KR1020110077912
    • 2011-08-04
    • 엘지전자 주식회사
    • 김종헌박영경이중재김현수박준오박안드레아스셰이커디르산드라이제훈
    • B60W30/12B60W40/02G06T7/60
    • B60W40/072B60W40/06B60W2420/42B60W2520/10B60W2540/18G06T7/12G06T7/174G06T7/181G06T2207/10012G06T2207/10016G06T2207/30256
    • 본명세서는영상을획득하는카메라모듈; 상기획득된영상으로부터복수의특징점을추출하고, 상기추출된복수의특징점을하나의선으로연결하는차선적합을수행하고, 상기적합된 차선을추적하는제어부; 및상기추적된차선을표시하는표시부를포함하되, 상기차선적합은, 임의의과거시점에서추출된특징점정보를획득하고, 상기임의의과거시점으로부터현재까지의주행정보를기초로상기과거시점에서추출되었던특징점에대한현재시점의위치를추정하고, 상기추정된현재시점의위치를기초로차선의횡방향치우침을나타내는오프셋(offset)을결정하고, 상기결정된오프셋을기초로곡선적합을수행하는것을특징으로하는차선인식장치에관한것이다. 또한, 본명세서는영상을획득하는단계; 상기획득된영상으로부터복수의특징점을추출하는단계; 상기추출된복수의특징점을하나의선으로연결하는차선적합을수행하는단계; 상기적합된 차선을추적하는단계; 및상기추적된차선을표시하는단계를포함하되, 상기차선적합을수행하는단계는, 임의의과거시점에서추출된특징점정보를획득하는단계; 상기임의의과거시점으로부터현재까지의주행정보를기초로상기과거시점에서추출되었던특징점에대한현재시점의위치를추정하는단계; 상기추정된현재시점의위치를기초로차선의횡방향치우침을나타내는오프셋(offset)을결정하는단계; 및상기결정된오프셋을기초로곡선적합을수행하는단계를포함하는것을특징으로하는차선인식방법에관한것이다.
    • 公开了一种用于检测车道的装置,该装置包括:相机模块,其捕获图像; 控制单元,从所述图像中提取多个特征点,进行车道拟合以将所述多个特征点与单线连接,并跟踪所述车道的装配; 以及显示单元,其显示被跟踪的车道,其中,所述车道配合包括获取关于在任意过去时间提取的特征点的信息,基于来自当前时间的驾驶信息来估计当前时间在过去时提取的特征点的位置 根据当前时间的任意过去时间,基于当前时间的位置确定表示车道的横向倾斜的偏移,并且基于偏移量进行曲线拟合。 此外,公开了一种检测车道的方法,该方法包括:拍摄图像; 从图像中提取多个特征点; 进行车道装配以将多个特征点与单个线路连接; 跟踪车道配合; 以及显示被跟踪的车道,其中,所述车道配合的执行包括获取关于在任意过去时间提取的特征点的信息; 基于从任意的过去时间到当前时间的驾驶信息,在当前时间估计在任意过去时间提取的特征点的位置; 基于当前时间的位置确定表示车道横向倾斜的偏移量; 并基于偏移量进行曲线拟合。
    • 9. 发明公开
    • 결함 검사 방법
    • 缺陷检查方法
    • KR1020140147685A
    • 2014-12-30
    • KR1020140067399
    • 2014-06-03
    • 가부시키가이샤 도쿄 웰드
    • 요코야마요시히코
    • G01N21/88G01B11/14
    • G06T7/0085G06K9/4638G06T7/0004G06T7/13G06T7/155G06T7/181
    • The present invention relates to a defect inspection method by which defects such as cracking can be detected with high precision. The present invention includes a process for obtaining a captured image. A high evaluation value is assigned, through scanning of the captured image in a plurality of directions, in a case where the brightness of each pixel M is lower than those of first adjacent pixels K and 0 on both sides of the scanning direction with regard to the scanning directions and the brightness of second adjacent pixels C and W on both sides orthogonal to the scanning direction is lower than those of third adjacent pixels A and E and U and Y on both sides of the scanning direction. A selected pixel for each scanning direction is selected based on each pixel evaluation value, and then the selected pixels are connected for each scanning direction. The connected pixels that deviate from a predetermined shape are removed.
    • 本发明涉及可以高精度地检测出裂纹等缺陷的缺陷检查方法。 本发明包括获得拍摄图像的处理。 在每个像素M的亮度低于扫描方向两侧的第一相邻像素K和0的情况下,通过扫描多个方向上的拍摄图像来分配高评价值 与扫描方向正交的两侧的第二相邻像素C和W的扫描方向和亮度比扫描方向两侧的第三相邻像素A,E以及U,Y的亮度低。 基于每个像素评估值选择用于每个扫描方向的所选像素,然后针对每个扫描方向连接所选择的像素。 去除了偏离预定形状的连接像素。