CN101031199A

CN101031199A - Linear form and position adapting method based on pin

Info

Publication number: CN101031199A
Application number: CN 200710072873
Authority: CN
Inventors: 钟江生; 廖强华; 李秦川; 郭树军; 夏毓鹏
Original assignee: Shenzhen Polytechnic
Current assignee: Shenzhen Polytechnic
Priority date: 2007-01-18
Filing date: 2007-01-18
Publication date: 2007-09-05

Abstract

The invention relates to a pin-based shape and position straight line fitting detection method, which comprises the steps of: 1) setting a region of interest ROI; 2) image processing; 3) rough positioning; 4) component feature detection; 5) precise positioning, In the second step, the present invention uses the improved straight line fitting method to carry out straight line fitting on the center points of each pin of the component, which can remove the deviation points and make the straight line fitting effect better. The accuracy and speed requirements of the machine component positioning detection system can accurately and efficiently detect and identify components during the placement process.

Description

Fitting method of shape and position line based on pin

技术领域technical field

本发明属于表面贴装工艺中的贴片机视觉系统使用的图像处理领域，尤其是指一种基于引脚的形位直线拟合方法。The invention belongs to the field of image processing used by the visual system of a placement machine in a surface mounting process, and in particular refers to a pin-based shape and position straight line fitting method.

背景技术Background technique

视觉对中检测系统是通过摄像机利用图像处理技术对待贴装的元件进行封装检测及中心和转角的检测，是贴片机能够快速准确地完成贴装任务的关键技术之一。其主要难点在于飞行视觉(On-Fly-Vision)即图像处理的实时性，并且待贴装元件的封装形式多样，从微型、超小型元件，直到大型的封装器件，每一种封装类型的检测算法都不一样，尤其是一些大型的封装如QFP(Quad FlatPackage，四周扁平封装)，BGA(Ball Grid Array，阵列锡球封装)等，而细间距SMD(Surface Mount Device，表面贴装元件)其检测算法更是难点，而这方面的技术历来受到专利保护，在国内更是鲜为人知。高精度机器视觉定位系统是满足贴装精度的有效途径。The visual alignment detection system uses the camera to use the image processing technology to perform package detection and detection of the center and corner of the component to be mounted. It is one of the key technologies for the placement machine to quickly and accurately complete the placement task. The main difficulty lies in the real-time nature of On-Fly-Vision (On-Fly-Vision), that is, image processing, and the various packaging forms of the components to be mounted, from micro, ultra-small components to large packaged devices, the detection of each package type Algorithms are different, especially some large packages such as QFP (Quad Flat Package, flat package around), BGA (Ball Grid Array, array solder ball package), etc., and fine-pitch SMD (Surface Mount Device, surface mount components) The detection algorithm is even more difficult, and the technology in this area has always been protected by patents, and it is little known in China. High-precision machine vision positioning system is an effective way to meet the placement accuracy.

发明内容Contents of the invention

本发明所要解决的技术问题在于：提供一种基于引脚的形位直线拟合方法，其可准确、高效地在贴装过程中进行元器件的检测与识别。The technical problem to be solved by the present invention is to provide a shape-position straight-line fitting method based on pins, which can accurately and efficiently detect and identify components during the mounting process.

本发明提供一种基于引脚的形位直线拟合方法，其包括以下步骤：The invention provides a pin-based shape and position straight line fitting method, which comprises the following steps:

第一步、设置感兴趣区域ROI步骤，将包含芯片所在区域设置为感兴趣区域ROI，并对该感兴趣区域ROI图像进行二值化；The first step, setting the region of interest ROI step, setting the region containing the chip as the region of interest ROI, and binarizing the region of interest ROI image;

第二步、图像处理步骤，对上述感兴趣区域ROI进行预处理，从中分割出芯片的各个引脚区域，判断引脚数目是否符合该芯片应有的引脚数目；The second step, the image processing step, preprocesses the ROI of the above-mentioned region of interest, divides each pin area of the chip therefrom, and determines whether the number of pins meets the number of pins that the chip should have;

第三步、粗定位，利用芯片各个引脚的中心位置初步确定元件的中心坐标(s_x′，s_y′)，对芯片的引脚中心点进行直线拟合，得到四条直线，计算此四条直线与水平轴偏差的角度，取平均值为θ′，并将其中最大的角度差记为β；The third step is rough positioning. Use the center position of each pin of the chip to preliminarily determine the center coordinates (s _x ′, s _y ′) of the component, and perform straight line fitting on the center point of the pins of the chip to obtain four straight lines. Calculate these four The angle of deviation between the straight line and the horizontal axis, the average value is θ', and the largest angle difference is recorded as β;

第四步、元件特征检测，在未二值化的图像上放置标尺对元件引脚进行测量，得到包括引脚宽、引脚长以及引脚的相对位置信息，将结果与标准参数进行比较，判断元件是否合格，若合格则进入第五步；The fourth step is component feature detection. Place a ruler on the non-binarized image to measure the component pins, and obtain information including pin width, pin length, and relative position of the pins, and compare the results with standard parameters. Determine whether the component is qualified, and if it is qualified, enter the fifth step;

第五步、精确定位，设精确转角为θ，其在(θ′-β/2，θ′+β/2)范围内，精确中心(s_x，s_y)在以(s_x′，s_y′)为中心，δ为半径的范围之内，δ为第三步中粗定位结果的最大可能误差，可根据经验设定，设图像分割中各计算出的元件引脚中心为a_i，由先验知识可以得到待贴元件各对应引脚的中心坐标为c_i，对于每对中心点之间的误差向量e_i由下式求出：The fifth step, accurate positioning, set the precise rotation angle as θ, which is within the range of (θ′-β/2, θ′+β/2), and the precise center (s _x , s _y ) is at (s _x ′, s _y ′) is the center, δ is within the range of the radius, and δ is the maximum possible error of the rough positioning result in the third step, which can be set according to experience. Let the center of each component pin calculated in the image segmentation be a _i , From the prior knowledge, the center coordinates of the corresponding pins of the components to be attached can be obtained as c _i , and the error vector e _i between each pair of center points can be obtained by the following formula:

${e e}_{i i} = = [\begin{matrix} {s the s}_{x x} \\ {s the s}_{y the y} \end{matrix}] + + [\begin{matrix} cos cos θ θ & - - sin sin θ θ \\ sin sin θ θ & cos cos θ θ \end{matrix}] [\begin{matrix} {x x}_{{a a}_{i i}} \\ {y the y}_{{a a}_{i i}} \end{matrix}] - - [\begin{matrix} {x x}_{{c c}_{i i}} \\ {y the y}_{{c c}_{i i}} \end{matrix}] = = [\begin{matrix} 11 & 00 & {x x}_{{a a}_{i i}} & - - {y the y}_{{a a}_{i i}} \\ 00 & 11 & {y the y}_{{a a}_{i i}} & {x x}_{{a a}_{i i}} \end{matrix}] [\begin{matrix} {s the s}_{x x} \\ {s the s}_{y the y} \\ cos cos θ θ \\ sin sin θ θ \end{matrix}] - - [\begin{matrix} {x x}_{{c c}_{i i}} \\ {y the y}_{{c c}_{i i}} \end{matrix}] - - - - - - ((11))$

设：set up:

$\{\begin{matrix} {s the s}_{x x} = = {s the s}_{x x}^{' '} + + k k * * Δx Δx & k k < < δ δ / / Δx Δx \\ {s the s}_{y the y} = = {s the s}_{y the y}^{' '} + + l l * * Δy Δy & l l < < δ δ / / Δy Δy \\ θ θ = = {θ θ}^{' '} + + h h * * Δθ Δθ & h h < < β β / / Δθ Δθ \end{matrix} - - - - - - ((22))$

进行搜索求使得∑e_i ²取值最小时的k，l，h的值，根据(2)式可得出精确定位结果(S_x，S_y，θ)。Search for the values of k, l, h when ∑e _i ² takes the minimum value, and the precise positioning results (S _x , S _y , θ) can be obtained according to formula (2).

其中所述第三步中对X引脚集和Y引脚集的中心坐标进行直线拟和的方法包括以下步骤：Wherein the method for carrying out linear fitting to the central coordinates of the X pin set and the Y pin set in the third step includes the following steps:

1)对X引脚集和Y引脚集中所包含的所有引脚的中心坐标点进行最小二乘法直线拟合，公式为Y＝a′*X+b′；1) Carry out the least square method straight line fitting to the central coordinate points of all pins contained in the X pin set and the Y pin set, the formula is Y=a'*X+b';

2)判断上述中心坐标点中是否大于所设定的拟合直线所需的最小点数N_min，如果否，则退出；如果是，则求出每个中心坐标点到直线的距离z_i，并判断其中的最大值z_k是否大于给定的阈值z_max，如果是，则剔除该中心坐标点，然后对剩下的N-1个中心坐标点进行最小二乘法拟合，直到第l次与第l-1次拟合的直线的斜率差不大于给定的斜率误差ξ时结束；2) Judging whether the above-mentioned central coordinate points are greater than the set minimum number of points N _min required for fitting the straight line, if not, then exit; if yes, then find the distance z _i from each central coordinate point to the straight line, and Judging whether the maximum value z _k is greater than the given threshold z _max , if yes, remove the central coordinate point, and then perform least squares fitting on the remaining N-1 central coordinate points until the lth time with End when the slope difference of the l-1th fitted straight line is not greater than the given slope error ξ;

3)最后得出结果Y＝a*X+b。3) Finally, the result Y=a*X+b is obtained.

本发明的有益效果是：本发明能够满足贴片机元件定位检测系统的精度和速度要求，可准确、高效地在贴装过程中进行元器件的检测与识别。The beneficial effects of the present invention are: the present invention can meet the accuracy and speed requirements of the component positioning detection system of the placement machine, and can accurately and efficiently detect and identify components during the mounting process.

附图说明Description of drawings

图1为本发明设置的感兴趣区域的示意图；Fig. 1 is the schematic diagram of the region of interest provided by the present invention;

图2为本发明定义的元件引脚集示意图；Fig. 2 is a schematic diagram of component pin sets defined by the present invention;

图3为本发明方法的流程图；Fig. 3 is the flowchart of the inventive method;

以下结合附图对本发明做进一步说明。The present invention will be further described below in conjunction with the accompanying drawings.

具体实施方式Detailed ways

本发明的基于引脚的形位直线拟合方法包括以下步骤：The shape and position straight line fitting method based on the pin of the present invention comprises the following steps:

第一步、设置感兴趣区域ROI，如图1所示，在芯片图像处理过程中，由于芯片大小各不相同，在多数情况下芯片在图像中所占的区域面积并不是很大，若对整幅图像进行分析处理，会浪费大量时间。为了提高检测算法的效率，在处理过程中，根据送料器的位置和规划出来的贴片头的运动轨迹以及元件的大小等信息，可以大致确定一个芯片图像所处的区域，忽略图像中没有包含芯片特征的区域，将仅包含芯片所在的区域设置为感兴趣区域ROI，并进行二值化。The first step is to set the region of interest ROI, as shown in Figure 1, in the chip image processing process, because the size of the chip is different, in most cases the area occupied by the chip in the image is not very large. Analyzing and processing the entire image will waste a lot of time. In order to improve the efficiency of the detection algorithm, in the processing process, according to the position of the feeder, the planned trajectory of the placement head and the size of the component, the area where a chip image is located can be roughly determined, ignoring that the image does not contain chips. For the feature region, set only the region where the chip is located as the region of interest ROI, and perform binarization.

第二步、图像处理，对感兴趣区域ROI进行预处理，并从中分割出元件的各个引脚区域。考虑元件的外形参数是已知的，可以确定出对应于引脚区域的面积应在特定范围内。利用这一先验知识，对标识后的区域计算面积特征和方向特征。如果该区域的面积不在设定的范围之内，将其设为背景，则剩下的区域即对应元件的各个引脚，并通过元件的中心和引脚区域的方向可以将其分为某个边上的引脚。至此可以通过区域的数目来判定元件是否缺脚，如果区域数目与元件应有的引脚数目不符合，则认为元件不合格，重新拾取下一元件。The second step, image processing, preprocesses the region of interest ROI, and segments each pin region of the component from it. Considering that the shape parameters of the component are known, it can be determined that the area corresponding to the lead area should be within a certain range. Using this prior knowledge, the area feature and direction feature are calculated for the marked area. If the area of the area is not within the set range, set it as the background, then the remaining area corresponds to each pin of the component, and it can be divided into a certain area by the center of the component and the direction of the pin area side pins. At this point, the number of areas can be used to determine whether the component is missing pins. If the number of areas does not match the number of pins that the component should have, the component is considered unqualified and the next component is picked up again.

定义元件的四个边为：NOUTH、SOUTH、EAST、WEST，如图2所示，对应的四个引脚集：NLeadset、SLeadset、ELeadset、WLeadset。Define the four sides of the component as: NOUTH, SOUTH, EAST, and WEST, as shown in Figure 2, and the corresponding four pin sets: NLeadset, SLeadset, ELeadset, WLeadset.

根据引脚的中心坐标和引脚的外形信息来确定引脚归属于哪个引脚集，由于元件引脚一般具有细长型的特点，这样就一个方向的特征可用，也就是说NLeadset、SLeadset引脚集的引脚的方向和Y方向的夹角比较小，而与X方向的夹角就比较大，将其称为Y方向的引脚；同理，ELeadset、WLeadset引脚集的引脚和X方向的夹角比较小，而与Y方向的夹角就比较大，将其称为X方向的引脚。通过元件引脚的方向将其分为Y方向引脚和X方向引脚，然后再根据每个引脚中心坐标在整个元件中心坐标的方位确定出每个元件引脚的确切归属。Determine which pin set the pin belongs to according to the center coordinates of the pin and the shape information of the pin. Since the component pins generally have the characteristics of slender shape, the characteristics of one direction are available, that is to say, NLeadset and SLeadset lead The angle between the pin direction of the pin set and the Y direction is relatively small, and the angle between the pin set and the X direction is relatively large, which is called the pin in the Y direction; similarly, the pins of the ELeadset and WLeadset pin sets and The included angle in the X direction is relatively small, while the included angle with the Y direction is relatively large, which is called the pin in the X direction. According to the direction of the component pins, it is divided into Y-direction pins and X-direction pins, and then the exact ownership of each component pin is determined according to the orientation of the center coordinates of each pin in the center coordinates of the entire component.

第三步、粗定位，利用各个引脚的中心位置初步确定元件的中心(s_x′，s_y′)，对四个引脚集的中心点进行直线拟和，可以得到四条直线NLine、SLine、ELine、WLine。并且将直线ELine、WLine旋转90度，旋转后理论上这四条直线应该是平行的，但是实际上拟和出的直线并不是完全平行的，一般情况下这四条直线相对于水平轴的角度是不相等的，其必然存在一个角度差作为粗定位可对其取平均值，结果记为θ′。将其最大角度差记为β。The third step is rough positioning. Use the center position of each pin to preliminarily determine the center of the component (s _x ′, s _y ′), and perform straight line fitting on the center points of the four pin sets to obtain four straight lines NLine and SLine , ELine, WLine. And rotate the straight lines ELine and WLine by 90 degrees. After the rotation, the four straight lines should be parallel in theory, but in fact the fitted straight lines are not completely parallel. Generally, the angles of the four straight lines relative to the horizontal axis are different. Equal, there must be an angle difference as the rough positioning can be averaged, and the result is recorded as θ'. Denote the maximum angle difference as β.

第四步、元件特征检测，在未二值化的图像上放置标尺对元件引脚进行测量，包括引脚宽、引脚长以及引脚的相对位置等信息，将结果与标准参数进行比较，判断元件是否合格。The fourth step, component feature detection, place a ruler on the unbinarized image to measure the component pins, including pin width, pin length, and relative position of the pins, and compare the results with standard parameters. Determine whether the component is qualified.

第五步、精确定位，粗定位的结果为(s_x′，s_y′，θ′)，误差较大，从第二步的结果知道转角θ应该在(θ′-β/2，θ′+β/2)范围内，同理，元件的精确中心(s_x，s_y)应该在以(s_x′-s_y′)为中心，δ为半径的范围之内，δ为第三步中粗定位结果的最大可能误差，可根据经验设定，设图像分割中各计算出的元件引脚中心为a_i，由先验知识可以得到待贴元件各对应引脚的中心坐标为c_i，对于每对中心点之间的误差向量e_i由下式求出：The fifth step is precise positioning. The rough positioning result is (s _x ′, s _y ′, θ′), and the error is relatively large. From the result of the second step, we know that the rotation angle θ should be in (θ′-β/2, θ′ +β/2), similarly, the exact center of the component (s _x , s _y ) should be within the range centered on (s _x ′-s _y ′), and δ is the radius, and δ is the third step The maximum possible error of the medium and coarse positioning results can be set according to experience. Let the center of each component pin calculated in image segmentation be a _i , and the center coordinates of each corresponding pin of the component to be pasted can be obtained from prior knowledge as c _i , the error vector e _i between each pair of center points is obtained by the following formula:

${e e}_{i i} = = [\begin{matrix} {s the s}_{x x} \\ {s the s}_{y the y} \end{matrix}] + + [\begin{matrix} cos cos θ θ & - - sin sin θ θ \\ sin sin θ θ & cos cos θ θ \end{matrix}] [\begin{matrix} {x x}_{{a a}_{i i}} \\ {y the y}_{{a a}_{i i}} \end{matrix}] - - [\begin{matrix} {x x}_{{c c}_{i i}} \\ {y the y}_{{c c}_{i i}} \end{matrix}] = = [\begin{matrix} 11 & 00 & {x x}_{{a a}_{i i}} & - - {y the y}_{{a a}_{i i}} \\ 00 & 11 & {y the y}_{{a a}_{i i}} & {x x}_{{a a}_{i i}} \end{matrix}] [\begin{matrix} {s the s}_{x x} \\ {s the s}_{y the y} \\ cos cos θ θ \\ sin sin θ θ \end{matrix}] - - [\begin{matrix} {x x}_{{c c}_{i i}} \\ {y the y}_{{c c}_{i i}} \end{matrix}]$

设：set up:

$\{\begin{matrix} {s the s}_{x x} = = {s the s}_{x x}^{' '} + + k k * * Δx Δx & k k < < δ δ / / Δx Δx \\ {s the s}_{y the y} = = {s the s}_{y the y}^{' '} + + l l * * Δy Δy & l l < < δ δ / / Δy Δy \\ θ θ = = {θ θ}^{' '} + + h h * * Δθ Δθ & h h < < β β / / Δθ Δθ \end{matrix}$

进行搜索求使得∑e_i ²取值最小时的k，l，h的值，根据上式即可精确定位元件。Search for the values of k, l, and h when ∑e _i ² takes the smallest value, and the components can be precisely positioned according to the above formula.

本发明方法的流程图如图3所示。The flowchart of the method of the present invention is shown in FIG. 3 .

另外，本发明对现有的直线拟合方法进行了改进，现在通常使用的直线拟合方法一般是最小二乘法。这种方法的缺点是，若参加直线拟合的部分点出现偏差，将会影响整个拟合效果。考虑到光照以及元件本身可能出现的引脚翘曲，个别点可能会不属于该直线，因此该点不应该参加直线拟合。传统的最小二乘法难以去掉这些点。为此本发明设计了一种更稳健的直线拟合算法，它能很好地解决这一问题。其基本思想是：In addition, the present invention improves the existing straight line fitting method, and the currently commonly used straight line fitting method is generally the least square method. The disadvantage of this method is that if some of the points participating in the straight line fitting deviate, it will affect the entire fitting effect. Considering the lighting and the possible lead warpage of the component itself, individual points may not belong to the straight line, so this point should not participate in the straight line fitting. The traditional least squares method is difficult to remove these points. For this reason, the present invention designs a more robust straight line fitting algorithm, which can solve this problem well. The basic idea is:

现根据N个点利用最小二乘法拟合得出直线L′，其直线方程为Y＝a′*X+b′；再判断参与拟合直线的点是否大于所设定的拟合直线所需的最小点数N_min，如果否，则退出；如果是，则求出每个点到直线L′的距离z_i，并判断其中的最大值z_k是否大于给定的阈值z_max，如果是，则说明该点不属于该直线，剔除该点，然后对剩下的N-1个点进行最小二乘法拟合，直到第l次与第l-1次拟合的直线的斜率不大于给定的斜率误差ξ时结束，最后得出精确直线L，直线方程为Y＝a*X+b。Now according to the N points, use the least square method to fit the straight line L', and the straight line equation is Y=a'*X+b'; then judge whether the points participating in the fitting straight line are larger than the set fitting straight line required The minimum number of points N _min , if not, exit; if yes, find the distance _zi from each point to the line L′, and judge whether the maximum value z _k is greater than the given threshold z _max , if yes, Then it means that the point does not belong to the straight line, remove the point, and then perform least squares fitting on the remaining N-1 points until the slope of the line fitted between the lth and l-1th times is not greater than the given It ends when the slope error ξ is higher, and finally an accurate straight line L is obtained, and the equation of the straight line is Y=a*X+b.

将此改进行直线拟合方法运用于本发明的第三步，元线各引脚中心点的直线拟合时，则同理，将各中心点坐标先进行最小二乘法直线拟合，再对各中心点到预设的直线的距离进行循环判断，踢除最大距离的点，再对剩下的中心点进行最小二乘法拟合，如此循环，直到第l次与第l-1次拟合的直线的斜率不大于给定的斜率误差ξ时结束，最后可精确得出X引脚集和Y引脚集的其四条直线，NLine、SLine、ELine、WLine。This change is carried out straight line fitting method and is applied to the 3rd step of the present invention, when the straight line fitting of each pin center point of element line, then in like manner, each center point coordinate is first carried out least square method straight line fitting, then to The distance from each center point to the preset straight line is cyclically judged, the point with the largest distance is kicked out, and then the remaining center points are fitted by the least square method, and so on until the lth and l-1th fitting When the slope of the straight line is not greater than the given slope error ξ, the four straight lines of the X pin set and the Y pin set can be accurately obtained, NLine, SLine, ELine, WLine.

由此可见，本发明通过设置感兴趣区域ROI步骤、图像处理步骤、粗定位步骤、元件特征检测步骤和精确定位步骤，对芯片进行逐步检测是否合格，同时可精确定位芯片位置，还采用了改进的直线拟合方法，使粗定位步骤中的直线拟合结果更为精确，从而更进一步使精确定位步骤得出的定位结果更精确。It can be seen that the present invention can step by step check whether the chip is qualified by setting the region of interest ROI step, image processing step, rough positioning step, component feature detection step and precise positioning step, and can accurately position the chip position at the same time. The linear fitting method of the method makes the linear fitting result in the coarse positioning step more accurate, thereby further making the positioning result obtained in the precise positioning step more accurate.

Claims

1, a kind of shape and position straight line fitting method based on pin, it comprises steps:

The first step, setting the region of interest ROI step, setting the region containing the chip as the region of interest ROI, and binarizing the region of interest ROI image;

The second step, the image processing step, preprocesses the ROI of the above-mentioned region of interest, divides each pin area of the chip therefrom, and determines whether the number of pins meets the number of pins that the chip should have;

The third step is rough positioning. Use the center position of each pin of the chip to preliminarily determine the center coordinates (s _x ′, s _y ′) of the component, and perform straight line fitting on the center point of the pins of the chip to obtain four straight lines. Calculate these four The angle of deviation between the straight line and the horizontal axis, the average value is θ', and the largest angle difference is recorded as β;

The fourth step, component feature detection, place a ruler on the non-binarized image to measure the component pins, and obtain information including pin width, pin length, and relative position of the pins, and compare the results with standard parameters. Judging whether the component is qualified, if qualified, enter the fifth step;

The fifth step, accurate positioning, set the precise rotation angle as θ, which is within the range of (θ′-β/2, θ′+β/2), and the precise center (s _x , s _y ) is at (s _x ′, s _y ′) is the center, δ is within the range of the radius, and δ is the maximum possible error of the rough positioning result of the third step above. Let the center of each component pin calculated in the image segmentation be a _i , which can be obtained from the prior knowledge The center coordinates of the corresponding pins of the components to be pasted are c _i , and the error vector e _i between each pair of center points is obtained by formula (1):

{e e}_{i i} = = [\begin{matrix} {s the s}_{x x} \\ {s the s}_{y the y} \end{matrix}] + + [\begin{matrix} cos cos θ θ & - - sin sin θ θ \\ sin sin θ θ & cos cos θ θ \end{matrix}] [\begin{matrix} {x x}_{{a a}_{i i}} \\ {y the y}_{{a a}_{i i}} \end{matrix}] = = [\begin{matrix} 11 & 00 & {x x}_{{a a}_{i i}} & - - {y the y}_{{a a}_{i i}} \\ 00 & 11 & {y the y}_{{a a}_{i i}} & {x x}_{{a a}_{i i}} \end{matrix}] [\begin{matrix} {s the s}_{x x} \\ {s the s}_{y the y} \\ cos cos θ θ \\ sin sin θ θ \end{matrix}] - - [\begin{matrix} {x x}_{{c c}_{i i}} \\ {y the y}_{{c c}_{i i}} \end{matrix}] - - - - - - ((11))

set up:

\{\begin{matrix} {s the s}_{x x} = = {s the s}_{x x}^{' '} + + k k * * Δx Δx & k k < < δ δ / / Δx Δx \\ {s the s}_{y the y} = = {s the s}_{y the y}^{' '} + + l l * * Δy Δy & l l < < δ δ / / Δy Δy \\ θ θ = = {θ θ}^{' '} + + h h * * Δθ Δθ & h h < < β β / / Δθ Δθ \end{matrix} - - - - - - ((22))

Search for the values of k, l, h when ∑e _i ² takes the minimum value, and the precise positioning results (S _x , S _y , θ) can be obtained according to formula (2).

2. The pin-based shape-line fitting method according to claim 1, characterized in that, the method for carrying out straight-line fitting to the center coordinates of each pin of the chip in the third step comprises the following steps:

1) Carry out least square method straight line fitting to the center point of all pins, obtain straight line L ', its straight line equation is Y=a'*X+b';

2) Judging whether the above-mentioned center point is greater than the set minimum number of points N _min required for fitting the straight line, if not, then exit; if yes, then find the distance Z _i from each center coordinate point to the straight line L′, and Judging whether the maximum value Z _k is greater than the given threshold Z _max , if yes, remove the central coordinate point, and then perform least squares fitting on the remaining N-1 central coordinate points until the lth time with End when the slope difference of the l-1th fitted straight line is not greater than the given slope error ξ;

3) Finally, an accurate result straight line L is obtained, and its straight line equation is Y=a*X+b.