JP2014508979A - Method and system for determining surface defects in a model of an object - Google Patents

Abstract

  Embodiments of the present invention disclose a method for determining surface defects in an object model generated from an original object model by simulation of a machining process. The method obtains a surface orientation and a rate of change of the orientation based on a normal vector with respect to the surface, and identifies a surface defect based on the rate of change and a threshold. The threshold is determined based on the machining process.

Description

  The present invention relates generally to simulation of machining processes, and more particularly to identifying surface defects on an object from a model of the object rendered during the simulation.

Numerically controlled machine milling Simulating the process of numerically controlled (NC) machining is fundamentally important in computer aided design (CAD) and computer aided manufacturing (CAM). During simulation, a computer model of the workpiece is edited using a computer representation of the NC machining tool and a set of NC machining tool movements to simulate the machining process. The workpiece model and tool representation can be visualized during simulation to detect potential collisions between parts such as the workpiece and tool holder, and after simulation, verify the final shape of the workpiece be able to.

  The final shape of the workpiece is influenced by the choice of tool and tool movement. Instructions for controlling these movements are typically generated from a graphical representation of the desired final shape of the workpiece using a computer-aided manufacturing (CAM) system. These movements are usually implemented using a numerically controlled programming language, also known as preparation code or G code. See standards RS274D and DIN 66025 / ISO 6983.

  The G code generated by the CAM system may not produce an exact replica of the desired shape. Furthermore, the NC tool motion is governed by the working part of the NC machining system, but its speed, range of motion, and acceleration and deceleration capabilities are limited. Thus, actual tool movement may not strictly follow NC machining instructions.

  The discrepancy between the final shape of the workpiece and the desired shape of the workpiece is very small and may be difficult to see. Depending on the situation, these discrepancies may result in undesirable grooved scratches on the final shaped surface of the workpiece, with a depth and width of approximately a few micrometers in size and a length of tens of micrometers. Chipping may occur.

  Typically, a set of NC machining instructions is tested by machining a test workpiece formed from a softer and less expensive material before machining the desired part. If the visual inspection of the test workpiece finds an undesirable discrepancy in the test workpiece, the NC machining instructions can be changed accordingly.

  However, this manual test is time consuming and expensive. The time to machine a single test workpiece may take approximately several hours and may need to be repeated several times before an acceptable set of NC machining instructions is obtained. It is therefore desirable to test for these discrepancies using computer-based simulation and rendering. Examples of computer based simulations are described in US patent application Ser. Nos. 12 / 495,588 and 12 / 468,607, which are incorporated herein by reference.

  A particularly important application of NC machining is the production of molds and dies. Molds and dies are produced in relatively small quantities by NC machining for later use in mass production. Thus, defects in the mold and die can undesirably move into the production part. Molds and dies are often used to form parts with smooth, slowly changing "free-form" surfaces with high quality, aerodynamic, tactile or aesthetic finishes . For example, modern toothbrushes are cast from plastic using a mold and have a complex free-form profile that provides both aesthetic and tactile benefits. Similarly, dies for punching car body panels have a smooth, free-form profile that can greatly affect aerodynamic resistance, and thus fuel economy, and the vehicle's aesthetic appeal to consumers.

  Dies for punching large parts with free-form outlines are milled using a tool with a size (several thousand millimeters) and a relatively small free-form surface (several millimeters). Due to both the need for instructions (often millions), production can be very time consuming. The same applies to injection molds for plastic parts. Molds are typically large to allow simultaneous production of many parts so as to improve manufacturing efficiency.

  The NC machining simulator can reproduce very small defects in very large simulated parts, for example parts in the thousands of millimeters, for example defects of tens to hundreds of microns. The problem of locating these small defects is a very difficult task. For example, a human operator must patiently search the entire simulated mold on a fine scale, which is time consuming and error prone.

  One method determines NC machining defects by analyzing machining tool paths. In particular, for a point on the path, a vector perpendicular to the plane containing the two vectors connecting the point and the previous and next points, respectively, is calculated. The direction of the normal vector with respect to one side of the plane is determined by the sign of the curvature of the path. In the case of a smooth plane, normal vectors from successive points should be approximately parallel with the same sign of curvature. However, the method only considers the relationship between successive points on the machining path and does not take into account defects caused by adjacent paths and / or adjacent areas on the surface of the machining tool. Limited to defects caused by local variations only.

  Therefore, there is a need to identify surface defects on the object from the model of the object rendered during simulation of the NC machining process.

  An object of the present invention is to provide a method for identifying defects on the surface of an object from a model of the object.

  It is a further object of the present invention to provide a method for highlighting defects to a user.

  It is a further object of the present invention to provide a method for highlighting surface defects in a model of an object rendered during simulation of a machining process.

  It is a further object of the present invention to provide a method for generating a list of possible defect areas on a simulated surface and presenting this list to the user.

  It is a further object of the present invention to provide a method that reduces memory requirements without compromising defect specific quality.

  Embodiments of the present invention are based on the recognition that the dependence of the apex angle on the cutting depth is reflected in the orientation of the surface of the object and / or the model of the object rendered during the simulation. For this reason, the defect of the model of an object can be judged using the direction of the surface, in particular, the change rate of the direction. For example, a less smooth region of the simulated surface can be identified and highlighted.

  The embodiment of the present invention determines a surface defect of an object based on the direction of the surface and the rate of change of the direction. In various embodiments, the rate of change is determined based on a normal vector to the surface, and surface defects are identified based on the rate of change and a threshold.

  For example, one embodiment is a method for determining defects in a surface of an object by a machining process applied to the object, wherein the surface is a model of the object by simulation of the machining process. The method comprises: determining the orientation of the surface at each pixel of the model of the surface; determining the rate of change of the orientation of the surface for each pixel of the model of the surface; Comparing the rate to at least one threshold to identify and compare the defects on the surface. The threshold value is determined based on the machining process.

  Another embodiment is a method for determining defects in a surface of an object by a machining process applied to the object, wherein the surface is generated from a model of the object by simulation of the machining process And determining a normal vector at each pixel of the surface and, for each pixel of the surface, based on a difference between a normal vector at the pixel and a normal vector at a neighboring pixel. Disclosed is a method that includes determining an orientation rate of change, wherein the surface defect is identified by the value of the rate of change and highlighting the defect on a display.

  Yet another embodiment is a system for determining surface defects in a model of an object, the means for generating the surface by simulation of the machining process, and in each pixel of the model of the surface, Means for determining the orientation of the surface based on a normal vector at a pixel; means for determining the rate of change of the orientation of the surface for each pixel of the model of the surface; and the rate of change at least one threshold value And a means for comparing to identify and compare the defects on the surface. The threshold value is determined based on the machining process.

FIG. 5 shows a ball end milling cutter used to produce a freeform surface during numerical control (NC) machining. It is a figure which shows the object milled by the ball end milling cutter of FIG. 1A. It is a side view of the object milled with the ball end milling cutter. It is a graph of the dependence of the angle of the cusp to the cutting depth. 2 is a block diagram of a method for determining a surface defect of an object according to one embodiment of the invention. FIG. 2 is a block diagram of a method for highlighting surface defects while rendering the surface of an object, according to one embodiment of the invention. FIG. FIG. 6 is a block diagram of an example for determining the rate of change of surface orientation, according to one embodiment of the present invention. FIG. 6 is an isometric view of an example model of an object with highlighted surface portions having defects. It is a block diagram of an example of one embodiment of the present invention.

  FIG. 1A shows a ball end milling cutter 100 that is commonly used to produce freeform surfaces during numerically controlled (NC) machining. In particular, the smooth hemispherical bottom surface 101 of the cutter 100 can machine a smooth surface.

  FIG. 1B shows an object 102 milled by three linear cutting sections 103, 104 and 105 of a ball end milling cutter 100 that produces a saddle-shaped surface. Between each cutting part, there is a triangular point formed by the intersection of each cutting part pair. For example, a sharp point 106 exists between the cutting parts 103 and 104.

  FIG. 2 is a side view of the object 201 milled by the four cutting parts 202 to 205 of the ball end milling cutter. Cutting parts 202, 203 and 205 have the same depth as indicated by dashed line 206. However, the cutting part 204 is deeper than the cutting parts 202, 203 and 205. Each cutting part pair intersects at the apex. For example, the cutting part 202 and the cutting part 203 intersect at the pointed part 207, and the cutting part 203 and the cutting part 204 intersect at the pointed part 208. Since the cutting portion 204 is deeper than the adjacent cutting portion, the angle 220 of the cusp 208 is larger than the angle 222 of the cusp 207. The distance between adjacent cuts is the step over distance 230 and is typically determined by the operator of the CAM system during the generation of machining instructions.

  FIG. 3 shows the dependence of the apex angle on the cutting depth. In this example, the angle of the cusp is obtained between two cutting parts separated by 0.2 mm. That is, the tool is moved in 0.2 mm incremental distance steps using a ball end mill cutting tool having a diameter of 4 mm. The angle increases as the cutting depth increases.

  Embodiments of the present invention are based on the recognition that the dependence of the apex angle on the cutting depth depends on the orientation of the surface of the object and / or the model of the object rendered during the simulation. For this reason, the defect of the model of an object can be judged using the direction of the surface, in particular, the change rate of the direction. For example, a rougher area of the simulated surface can be identified and highlighted.

  FIG. 4 shows a block diagram of a method 400 for determining surface defects in an object model. In various embodiments of the invention, the surface is the entire surface of the object, a portion of the entire surface of the object, a surface visible from a particular viewpoint, a region of the object with a high probability of defects, and / or combinations thereof. In one embodiment, the object model is generated from the original object model based on a set of machining instructions. The steps of the method are performed by a processor 401 known in the art.

  The object model surface 425 is simulated by the rendering module 420 based on the representation 410 underlying the object surface. The surface orientation module 430 determines the surface orientation 435 at each pixel of the surface.

  The orientation comparison module 440 determines a surface orientation rate of change 446 and compares it to one or more threshold values 445 to determine surface defects. One embodiment compares the rate of change with a minimum threshold. That is, the portion of the surface corresponding to the change rate of the surface orientation having a value higher than the minimum threshold is specified. Additionally or alternatively, another embodiment compares the rate of change with a maximum threshold. That is, the portion of the surface corresponding to the rate of change of the orientation of the surface having a value less than the maximum threshold is identified.

  In various embodiments, the minimum and / or maximum threshold is determined based on the original surface shape, tool size, machining instructions and / or desired accuracy.

  The minimum threshold facilitates the distinction between smooth region cusps over most of the surface of the simulated model of the object and actual defects in the surface that usually do not occur very often. As shown in FIG. 3, the angle of the apex is a function of the tool dimensions, eg diameter, and step over distance. For example, in one embodiment, the step over distance is selected based on the tool diameter and the desired smoothness of the surface, i.e., the height of the apex. For this reason, the desired angle of the apex is known before the machining process.

  The machining process includes, but is not limited to, turning operations, milling operations, and drilling operations. Depending on the operation, the machining process may select a machining tool, for example, select a tool type, tool shape, material and dimensions, such as a single point tool or a multi-blade tool, It further includes determining a direction, a tool step over distance, and determining a machining instruction for operation. The machining instructions include an operation sequence and a tool path.

  The machining process controls the desired accuracy of the surface. For example, the step over distance for rough cutting can be greater than the step over distance for finish cutting. On the other hand, the value of the step over distance is only an example of controlling the smoothness of the surface by a machining process. Thus, some embodiments of the present invention determine a minimum threshold and / or a maximum threshold based on a machining process that produces the surface of the object. Similarly, some embodiments determine the threshold based on a simulation of the machining process that produces the surface of the model of the object.

  For example, one embodiment determines the minimum threshold based on the desired angle value of the apex. One variation of this embodiment determines the minimum threshold by increasing the desired angle value of the apex by a margin that depends on the amount of allowable variation in the apex angle. .

  Additionally and alternatively, a minimum threshold can be determined based on the fact that surface defects are relatively rare. For example, one embodiment obtains a histogram of the rate of change of the number of pixels versus normal vector, described below, and selects the minimum threshold so that most pixels have a rate of change below the threshold. To do.

  The output 455 of the method is processed by the output module 450. For example, the output module stores the identified portion of the defective surface in a memory (not shown). Additionally or alternatively, the output module displays a model of the object on the display as shown in FIG. 7 and highlights the portion 701 of the surface that has a defect.

Normal Vector Some embodiments of the present invention determine the orientation of the surface based on the normal vector. As mentioned herein, the normal vector at a pixel has a length equal to 1 and is perpendicular to the surface at a location corresponding to the pixel. For example, referring to FIG. 2, vector 209 is a normal vector associated with the cutting portion 203 adjacent to the apex 208, and vector 210 is a normal vector associated with the cutting portion 204 adjacent to the apex 208. It is. Next, at each edge of the apex, the apex angle is determined by calculating the vector dot product of the surface normal vectors. The dot product is the cosine of the apex angle.

Online Processing FIG. 5 illustrates a method for highlighting surface defects in an object model while rendering the surface of the object. The rendering 502 of the simulated surface of the object relies on the underlying representation of the surface 501. For example, the surface can be represented by a Boolean difference between a distance field representing the original surface and a distance field representing the volume swept by the machining tool during cutting.

  In one embodiment, the surface is rendered by a conventional ray casting method. In the ray casting method, the mathematical ray associated with each pixel is projected onto the simulated surface from the viewing direction. The color and brightness of each ray that intersects the surface is determined by the surface color and normal vector at the intersection.

  Another representation 501 of the surface is a mesh of geometric primitives such as triangles. In one embodiment, the triangular mesh is rendered using a graphics application programming interface (API) such as OpenGL. Usually, color and normal vectors are defined at each vertex of a triangle interpolated by a graphic implementation across the triangle spanning pixels, according to viewing conditions.

  The normal vector is used to calculate the brightness of the pixel as if the surface were illuminated by light at a position above the surface. As an example, one conventional computer lighting technique uses pixel brightness as an environmental component unrelated to the normal vector, and a diffuse component proportional to the vector dot product of the normal vector and the vector from the surface to the illumination position. The sum of the surface normal vector and the specular component proportional to the dot product of the vector in the middle of the observation direction and the light direction is obtained as a power obtained by the glossiness of the surface.

  In conventional rendering, pixel brightness is calculated using the normal vector for each pixel in the surface image, and then the normal vector is discarded. The resulting pixel values are stored in a memory called a color frame buffer and are ultimately displayed, stored or transmitted.

  On the other hand, one embodiment of the present invention identifies and / or highlights surface defects instead of discarding the normal vector determined to calculate pixel brightness. To reuse. In this embodiment, instead of using the normal vector to instantly calculate pixel brightness, the normal vector 503 of all pixels in the surface image is stored in the normal frame buffer 504. Once all normal vectors are stored, the normal vectors are processed to identify and / or highlight defects.

  In one embodiment, the normal vector change rate 446 is obtained based on the center difference between the normal vectors. In particular, the rate of change in surface orientation at a pixel is determined as the difference between the normal vector at that pixel and the normal vector at neighboring pixels, eg, the normal vector at a pixel adjacent to that pixel.

  FIG. 6 shows the change in orientation of the surface at pixel 602, eg, center pixel 602 within window 601 based on the center difference between the normal vector at pixel 602 and the normal vector at adjacent pixels 602-610. An example for determining the rate is shown. Window 601 moves vertically (620) and / or horizontally (630) across all pixels of the surface model to determine the rate of change per pixel.

であり、８つの最も近傍のピクセル６０３〜６１０の法線ベクトルは、

である。ここで、ｉは、窓内の法線ベクトルのインデックスであり、例えば、１〜８の範囲をとり、ｘ、ｙおよびｚは軸ｘ、ｙおよびｚに沿った法線ベクトルの３次元（３Ｄ）成分を特定する。次に、ピクセルロケーション６０２における法線ベクトル成分の変化率Ｒ_０が、以下によって求められる。

ここで、ｊは、ベクトルの成分、すなわち、ｘ、ｙまたはｚを示す。法線ベクトルの変化率は、例えば、成分Ｒ_０ｊの二乗和平方根である。 Window 601 covers a 3 × 3 grid of normal vectors selected from the normal frame buffer. As defined herein, the normal vector at pixel 602 is

And the normal vectors of the eight nearest pixels 603-610 are

It is. Here, i is an index of a normal vector in the window, for example, having a range of 1 to 8, and x, y, and z are three-dimensional (3D) of the normal vector along the axes x, y, and z. ) Identify ingredients. Next, the rate of change R ₀ of the normal vector component at pixel location 602 is determined by:

Here, j represents a vector component, that is, x, y, or z. The rate of change of the normal vector is, for example, the square sum of squares of the component R _0j .

  Various embodiments of the present invention use different thresholds to identify surface defects. The area of the surface where the rate of change value is less than the minimum threshold 507 corresponds to a surface having a smooth surface and / or a cusp between non-defect cuts. A region of the surface having a rate of change value that exceeds a maximum threshold value 508 corresponds to a deep cut associated with an object edge and / or a non-freeform region of the surface. Thus, in one embodiment, only pixels that are within the defect range, i.e., corresponding to a value of the rate of change between the minimum and maximum thresholds, are identified as expected defect regions on the surface. .

  For example, one embodiment changes the color of the pixel corresponding to the rate of change value in the defect area to, for example, red (509) while the other pixel colors are stored in the normal frame buffer. The calculated value is calculated as usual (506). Another embodiment uses a lookup table to set the pixel color based on the rate of change. For example, pixels with a rate of change below the minimum threshold are marked blue, pixels with a rate of change within the defect range are marked red, and pixels with a rate of change above the maximum threshold are green Is attached. However, those skilled in the art will readily recognize that many other methods and / or look-up tables can be applied to improve the visual identification of defects. After pixels corresponding to regions having a rate of change within the defect range are determined, the pixels are stored in memory 511 and the color of these pixels is stored in color frame buffer 510.

Off-line processing Some embodiments identify surface defects in a process separate from the process of rendering and / or determining pixel brightness. These embodiments take into account that in some applications, the density of pixels is insufficient to detect small defects unless the surface is magnified during rendering. Similarly, insufficient pixel density can cause defects to be missed due to insufficient sampling of the surface.

  Additionally and / or alternatively, some surface defects are not visible from the viewing direction in the rendered model. In one embodiment, this limitation is overcome by observing the simulated surface on a dense scale over a wide angular range.

  On the other hand, another embodiment identifies surface defects after the simulation is complete, regardless of the view of the simulation results. Embodiments render portions of the surface in an off-screen normal buffer with a resolution high enough to ensure that defects larger than the minimum size can be identified. Additionally and / or alternatively, embodiments may inspect the simulated surface from a range of viewpoints, eg, along each of the + x, -x, + y, -y, + z and z axes. To do. Alternatively, one embodiment renders a portion of the surface rough, determines the dominant orientation of that portion of the surface, and renders the surface from that dominant orientation.

  In some embodiments, not all pixels with a rate of change of surface orientation between the minimum and maximum thresholds will correspond to surface defects. Some pixels may correspond to defects in the rendering process, such that the surface normal of the pixel is erroneously determined. Thus, one embodiment of the present invention first renders a portion of the surface rough and performs low resolution defect detection. Regions of the surface where defects are identified with low resolution are then locally re-rendered with high resolution to improve detection of defects identified by a row of pixels along the edge of the machined cut.

  One embodiment also uses a normal frame buffer to hold information regarding high resolution rendering of the surface. Alternative embodiments divide the surface into overlapping or adjacent patches, reducing memory requirements without compromising defect specific quality.

  Another embodiment generates a list of expected defect areas on the simulated surface and outputs this list to the user. For example, the list can be represented as a rectangular box overlaid on the low resolution image and surrounding the expected defect area. The user can then observe these areas more closely and make a final decision regarding the presence of actual defects. Additionally or alternatively, the defect list is presented in a text format that describes the expected defect location and characteristics. It is advantageous to use a user interface that allows the user to select the entry in the defect list to reorient, center, and zoom into the defect area of the simulated image.

  FIG. 8 shows an example of another embodiment of the present invention. A set of machining instructions 801 is provided to the NC machining controller 802 as a file over the network, from a CD or DVD, or by other means known in the art. The controller 802 includes a processor 803, a memory 804, and a display 805 for indicating the operation of the machine. The processor performs a machining simulation and executes a method according to an embodiment of the present invention, eg, method 400, to generate an image 507 on the display 505 that identifies a defect in the machining simulation.

  Although the invention has been described by way of examples of preferred embodiments, it is to be understood that various other adaptations and modifications can be made within the spirit and scope of the invention. Accordingly, it is the object of the appended claims to cover all variations and modifications falling within the true spirit and scope of the invention.

FIG. 8 shows an example of another embodiment of the present invention. A set of machining instructions 801 is provided to the NC machining controller 802 as a file over the network, from a CD or DVD, or by other means known in the art. The controller 802 includes a processor 803, a memory 804, and a display 805 for indicating the operation of the machine. The processor executes a machining simulation method according to an embodiment of the present invention, for example, to perform the method 400, the image 8 07 to identify the defects of machining simulation to generate on the display 8 05.

Claims (20)

A method for determining surface defects of a model of an object by a machining process applied to the object, wherein the surface is generated from the model of the object by simulation of the machining process, the method comprising:
Determining the orientation of the surface at each pixel of the model of the surface;
Determining the rate of change of the orientation of the surface for each pixel;
Comparing the rate of change with at least one threshold to identify and compare the defect on the surface, the threshold being determined based on the machining process A method for determining surface defects in a model of an object by a machining process applied to the object, wherein the steps of the method are performed by a processor.   The method of claim 1, further comprising determining the orientation of the surface based on a normal vector at each pixel.   The method of claim 1, further comprising determining the rate of change based on a difference between a normal vector at the pixel and a normal vector at a neighboring pixel.   The method of claim 3, wherein the neighboring pixel is directly adjacent to the pixel.

To determine the normal vector N ₀ , where N _0x , N _0y , N _0z are components of the normal vector along the axes x, y and z;

To determine the normal vector N _i according to where i is an index of the normal vector;

To determine the rate of change component R _0j according to: where j is x, y or z;
4. The method of claim 3, further comprising determining the rate of change based on the component of the rate of change. 6. The method of claim 5, further comprising determining the rate of change as a root sum square of the component _R0j . The comparison is
The method of claim 1, further comprising comparing the rate of change to a minimum threshold, wherein the pixels corresponding to the rate of change having a value exceeding the minimum threshold are identified. . The comparison is
Determining the minimum threshold based on a value of a cusp angle, the cusp further comprising being formed by a cutting part of a machining tool during simulation of the machining process; The method of claim 1. The comparison is
Comparing the rate of change to a minimum threshold;
Comparing the rate of change to a maximum threshold;
The method of claim 1, wherein the pixels corresponding to the rate of change that exceeds the minimum threshold and has a value less than the maximum threshold are identified. Rendering the surface of the model of the object with low resolution;
Determining a region of the surface identified by the defect;
Rendering the region of the surface with high resolution, thereby determining the defect of the surface within the region identified by a row of pixels along an edge of a machining cut The method of claim 1, further comprising: rendering with.   The method according to claim 1, wherein the object is formed by a cutting part of a milling cutter. Determining a normal vector at each pixel of the surface;
The method of claim 1, further comprising: storing the normal vector in a normal frame buffer.   The method of claim 1, further comprising highlighting a pixel corresponding to the defect. Changing the color of each pixel corresponding to the defect;
The method of claim 1, further comprising: rendering the model of the object on a display. A method for determining surface defects of a model of an object by a machining process applied to the object, wherein the surface is generated from the model of the object by simulation of the machining process, the method comprising:
Determining a normal vector at each pixel of the surface;
Determining, for each pixel on the surface, a rate of change in orientation of the surface based on a difference between a normal vector at the pixel and a normal vector at a neighboring pixel, wherein the surface defect is determined by A step identified by a value;
Highlighting the defect on a display. A method for determining a surface defect of a model of an object by a machining process applied to the object.   Determining a minimum threshold based on the machining process, the machining process further comprising at least one of a turning operation, a milling operation, and a drilling operation. Item 16. The method according to Item 15. Comparing the rate of change to a minimum threshold;
Comparing the rate of change to a maximum threshold value, wherein the pixel corresponding to the rate of change that exceeds the minimum threshold value and has a value less than the maximum threshold value is the defect on the surface. The method of claim 15, identified as the pixel corresponding to. A system for determining surface defects in a model of an object by a machining process applied to the object,
Means for generating the surface by simulation of the machining process;
Means for determining the orientation of the surface at each pixel of the model of the surface based on a normal vector at the pixel;
Means for determining a rate of change of the orientation of the surface for each pixel;
Means for comparing the rate of change with at least one threshold to identify and compare the defect on the surface, the threshold being determined based on the machining process A system for determining surface defects in a model of an object by a machining process applied to the object.   The system of claim 18, further comprising means for determining the rate of change based on a difference between the normal vector at the pixel and a normal vector at a neighboring pixel.   The system of claim 18, further comprising means for highlighting the pixels corresponding to the defect.

Images