DETAILED DESCRIPTION OF THE INVENTION

[0023] Universal Feature Object (UFO)—A Methodology to Store Features in a CAD Neutral Format

[0024] The present invention (which may be implemented as software, hardware, and/or a combination of software and hardware) includes a neutral file format that is capable of handling graphical representations of a 2D drawing and fully constrained feature-based 3D models from multiple CAD systems. This neutral or generic intermediate file format allows for the conversion of 2D legacy drawings to 3D feature-based models in a targeted CAD system and also allows for the collaborative update of a 3D model from one CAD system to another.

[0025] Referring now to FIG. 1 a, a block diagram of a three dimensional model 10 of the present invention is depicted. The model 10 consists of a collection of features 14 and may include a plurality of feature relationships. Each feature class contains the necessary information needed to build the feature including feature geometry 16 , constraints 18 , and dimensions 20 which are preferably saved at the feature level 14 and not at the model level 12 . The feature data is predominantly defined as a geometric representation of an individual feature type. This representation is preferably stored in a binary file format referred to as a Universal Feature Object (UFO). The order of features in the 3D model represents the order with which the model needs to be built in the targeted CAD system. It is the responsibility of the software associated with the individual CAD system interface to build the model in a manner consistent with that CAD system's expectations. The system interface software (named, for example, Feature Changer) is described in a subsequent section.

[0026] The UFO consists of two libraries that are necessary to create the 3D model. These two libraries are the geometry library and the feature library. The geometry library contains classes for all 2D and 3D entities as well as points and vectors. Common classes exist for 2D entities (for example, GE2D_Entity) as well as 3D entities (for example, GE3D_Entity). All entity types (lines, circles, etc.) will inherit from the common entity class. The data within each class is private and can only be accessed via member functions. In the geometry classes (both 2D and 3D), the member functions that set and retrieve the data actually copy the data between the class's private data space and the address of the data specified from the calling function. For example, a 2D line contains a start point. However, to retrieve this point, the calling function must pass an address where this particular point data is to be copied. This guarantees data integrity. Within each class, the data is classified as fundamental data, which is the minimum data necessary to define the entity, and the derived data, which is any information that may be used in or with the present invention. It is the responsibility of the individual classes to guarantee that any change made to the fundamental data via a function (for example, a member SET_function) will update the derived data accordingly. For example, if the start point of a line is changed via another function (for example, a Set_Start_Pt function), it is the responsibility of that function to update all fundamental data (i.e. length and slope of the line) accordingly.

[0027] An important class in the geometry library 16 is a 3D coordinate system class (for example, a GE3D_Coordinate_System class). The coordinate system is what permits each feature to be independently defined. It contains all the data necessary to detect a work plane, sketch plane, or face upon which a feature may need to be built. This data comprises plane vectors (positive x, positive y, and positive z), the origin of the plane, and the elevation (or distance) of the plane from world origin. In most cases, the geometry classes are simply a method of holding the data associated with some geometric type. An exception to this case is a class developed to aid in converting coordinate system data to other useful formats for use in the various applications (for example, a GE3D_Coordinate_System_Converter class).

[0028] Referring now to FIG. 1 b, a block diagram of common feature representations 30 is presented. All features typically inherit common properties from a generic, common, or base feature class (for example, FEAT_Base) 32 . This base feature class 32 stores data that is common across most feature types. Some common properties include feature type, the volume operation of the feature, and the feature constraints. The feature types include, for example, common features such as extrusion 34 , hole 36 , revolve 38 , chamfer 40 , fillet 42 , sweep 44 , shell 46 , loft 48 , as well as CAD specific features such as dome 50 . The volume operation consists of a join operation, a cut operation, and a volumetric intersection operation. The constraints are represented as an array of constraints and the constraint class is defined separately. Every individual feature 14 will typically inherit from this base class 32 . This allows for easy expandability as additional features are incorporated from future CAD systems. Each individual feature class contains feature-specific data. For example, the hole feature contains all the necessary data to create a hole (including specialized holes such as counter bore).

[0029] Several features are created by first drawing a profile shape from a collection of entities on a specific surface. Two such features are extrusion features and revolve features. A separate class is created to handle all profiles (for example, a FEAT_Profile class). The coordinate system needed to draw a feature is defined within the profile class and is due to the nature of the coordinate system class that contains the data necessary to detect the face or work plane. Within the profile class, the data needed to draw the profile can be handled as an array of 2D poly-lines or entities. Each CAD interface application can build the profile using the method that is native to that CAD system when drawing the profile.

[0030] Feature constraints are handled via two classes. The first class, (for example, a FEAT_Constraint class), defines the constraint type, constraint data value (i.e. dimension value) and up to two (or more) constraint objects (such as a 3D point or a 3D edge). The second class, (for example, a FEAT_Constraint_Object class), contains the constraint object type indicating the constraint is to an edge or to a point, and the point or edge definition.

[0031] In addition to containing a data representation for a 3D feature-based model, the UFO of the present invention also contains a method for storing the 2D views via a class (for example, a FEAT _— 2D_View class). The view class contains an array of 2D entities (lines, circles, arcs, etc.) and a coordinate system associated with the view. The view data is not necessary to build a 3D model and may be null.

[0032] By utilizing the current format for the UFO, the software of the present invention is easily maintained and updated. The UFO can be incrementally updated with additional features and little or no overhead. By using an object-oriented approach with the UFO, compatibility between versions is simplified. Upward compatibility between versions of the UFO can be achieved through the use of inheritance. Another compatibility issue deals with versions of a particular CAD system and involves attempting to transfer a model from a CAD system to an earlier version of that CAD system. With the exception of new features that are not handled in an earlier version, backward compatibility is achieved (using UFO) through the CAD application.

[0033] Another advantage of the UFO of the present invention is the amount of data transferred. The UFO file is a binary file of the individual CAD features and the metadata associated with each feature. By utilizing the application to determine the build method of the model, CAD specific data is not transferred through an intermediate file. Therefore, the UFO can be used in a collaborative environment. The UFO also lends itself to other possible uses such as an incremental update of a model in a collaborative environment. Finally, the UFO can handle manufacturing features as well as design features which include pocket, step, notch, etc.

[0034] A methodology for Incrementally Updating 3D Feature-Based CAD Models In a Collaborative Environment

[0035] 3D CAD models can get extremely large in size for complex parts and assemblies such as those used on airplanes and automobiles, for example. If such complex parts are to be collaboratively designed using conventional CAD systems, users have the choice of simply marking up a visualization version of the model (that is smaller in size) or transferring the model among them each time a design change is made. This is true of collaboration either in a homogeneous CAD environment where all users have the same CAD system or in a multi-CAD heterogeneous environment where all or some users do not have the same CAD system.

[0036] A feature of the present invention allows collaborators to only share design changes and not transfer entire product models between collaborators. This feature is a significant improvement over current collaboration models and enables users to collaborate on complex models over narrowband networks.

[0037] Referring now to FIG. 2 a, a system and flowchart for an incremental update procedure 60 is depicted. The system and flowchart 60 comprises a first client (client 1 ) 61 , a server that handles processing tasks 62 , and a second client (client 2 ) 63 . The first client 61 reads the features in the model that is currently open at step 64 and writes the model to a UFO format at step 65 . A comparison is performed with the older UFO (if one exists) at step 66 and any features that are changed are marked as shown at step 67 . The changed features are written to a separate class at step 68 and a unique ID is attached to the UFO at step 69 before sending the changes at step 70 to the server 62 . The server 62 receives the UFO from the first client 61 at step 71 , saves the UFO at step 72 and sends the UFO to the second client 63 at step 73 .

[0038] The second client 63 receives the changes from the server 62 at step 74 and compares the ID with the old UFO at step 75 . If the ID is new, the entire UFO is written to the old UFO at step 76 . Changes are updated in the old UFO at step 77 , otherwise the part is updated at step 78 . If the earlier collaboration was skipped, the entire part is redrawn at step 79 .

[0039] Referring now to FIG. 2 b, a flowchart 80 for incremental update using a Differential Universal Feature Object (DUFO) is presented. The collaboration is started at step 81 . All collaborating systems register with the server at step 82 . A system that needs to send a new part to other collaborators will open a new part at step 83 , attach a tag to the name of the part to identify it in collaboration at step 84 , and prepare to send the part for collaboration at step 85 . The client reads the part from the CAD system at step 86 which is checked to see if the part has a special tag attached to its name at step 89 . If there is a tag attached to its name, then the client searches for another file in the system that has the same tag in its name. If the UFO file is not found (at step 88 ) or there is no tag available for the part name, then a tag is added to the file and written out at step 94 . This new part is sent out for collaboration at step 95 and the system returns to step 85 and waits for collaborating on another part, or goes to step 90 to receive a new part from other systems.

[0040] If the file with the same tag is found at step 88 then this new file is read at step 96 . The new file is compared feature by feature with the old file at step 97 . Each of the features are individually compared at step 98 . If a match is found, the method moves on to the other features; if not, the new feature is marked as “Added” and is added to a new list at step 99 . This is done until the comparison is completed. When the comparison is completed, File 2 is compared with File 1 at step 100 . The unmatched feature is marked as “Deleted” and added to the new list at step 101 and the same process is continued. Once the comparison is complete, the features in the new list are compared at step 102 . If the features are the same, then the tag is changed to “Modified” and one copy is deleted from the list at step 104 .

[0041] Once the comparison is completed, a “DUFO” tag is added to the new list and is serialized at step 105 . This file is then sent to the server for collaboration at step 106 . The server then sends the file to all the other collaborators at step 107 . The other collaborating systems receive the file from the server at step 90 and check to see if there is a DUFO tag at step 91 . If there is a DUFO tag, the file is run through a design conversion engine to map the features at step 108 . The feature type and the index are received from the file, the corresponding feature is searched in the CAD system at step 109 , and the changes to the part in the CAD system are affected at step 110 . The same process is carried out on the saved UFO file and it is updated at step 111 . The CAD system waits for a new part or sends another part at steps 112 and 93 . When the collaborating system receives the part and if the part does not contain a DUFO tag, then the part is drawn on the CAD system at step 92 .

[0042] The following description further describes and clarifies the system and methods described in FIGS. 2 a and 2 b. At the start of the collaboration session, a first user either opens an existing model inside the CAD system or creates a new one. The DUFO file format has a tag associated with each file that indicates if the file has been used in collaboration. The first user should save the model as a UFO file. This initial UFO file does not have a tag attached to its name. When a model is saved as a UFO, it is checked to see if it has a unique tag attached to its name. If not, a unique tag is attached and the part is written out into a file, which has a unique tag as a part of its name. If the tag is still not present, the collaboration client knows that this part is new and simply writes it out into a file with that tag as part of the file name. The first client then sends the part to the server which distributes the file to all the collaborators.

[0043] At the receiving end, the collaboration client receives the file and checks to see if the file has a “DUFO” tag attached to it. This is done after serializing the UFO file. In the case where the entire part was sent out, no tag would be found and hence the entire part would have to rebuild/build on the receiving end CAD system. A unique tag that came with the UFO is added to the name of the file on the receiving CAD system. Now that the part has been sent to all collaborators, the users can collaboratively make design changes. The first collaborator can change the model and write it out to the UFO. Now the model has a unique tag attached to its name. This tag is taken and compared with all the UFO files that are in the current directory. A match may be found (for example, this UFO file can be referred to as file 2 ), and file 2 is serialized and compared with the feature tree that was read from the model. File 1 is compared with file 2 . Every feature from file 1 (the feature tree read from the model presently open) is sequentially compared with every feature in file 2 . If the feature is not found in file 2 , the feature is added to a new list as an “Added” feature. Once the comparison is complete, features from file 2 are compared with features from file 1 (for example, the feature tree that was read from the opened model). Any unmatched feature is added to the new list as a “Deleted” feature. When comparing, if the features are same but their feature index is different, then the feature is added to the tree as a “moved” feature.

[0044] Once this comparison is complete, features in the new list are compared with other features. If the feature type and the feature index (this is the index that keeps track of the relative position of the feature in the feature tree) are the same, one copy is deleted and the tag on the other one is changed to “Modified”. Once a complete comparison is performed, the new list is serialized (after adding a “DUFO” tag to the list). This new UFO is sent to the server which sends the file to all the collaborators other than the one that sent the file. The other collaborators receive the file and open it to see the DUFO tag. Each feature is extracted and is run through a feature-mapping engine to map the feature to an available feature in the CAD system. Once mapped, the mapped feature is taken and identified on the feature tree. The identification is accomplished using the feature type and the relative feature index in the tree. Once identified, the feature is either deleted, added or modified based on the tag that the feature carries. Added features are simply added without comparing these added features with existing features. Only modified and deleted features have to be compared.

[0045] In the case of fillets and chamfers, modified features are first deleted and then added. This is done because fillets and chamfers are typically dependant features and have a modified tag since the feature on which they are dependent is modified. This would make sure that the fillet and chamfer move to the right place when modified. When a fillet is deleted or added, it is reordered in the feature tree since all added features are by default added at the end of the feature tree. Since this does not maintain the design intent, when the feature is deleted, its position is noted down. Thus, when the modified feature is added, it is moved to the old position in the tree thereby maintaining the design intent.

[0046] Currently, when a feature is being modified, all attributes of the feature are modified even though a particular attribute would have remained the same. Eventually, the DUFO can note only the changes or mark and change the changed parameters. Once the feature in the DUFO file has been mapped and compared with the feature tree, the new features are added to the file that has the same tag as the part that is open in the CAD system. This makes sure that the UFO file in the system is up to date with the one opened in the CAD system. The feature is added or deleted from the list and its relative position is maintained in the list.

[0047] When collaborators are collaborating with a heterogeneous CAD system, a problem of un-mappable features usually arises. These features are either converted into another feature supported by the CAD system or added to the part as geometry (surfaces stitched together). In such a collaboration situation, incremental updating is difficult. For example, Collaborator A may have a dome feature and sends the file to Collaborator B who builds up the part by changing the dome to a loft in his/her system (since the dome is unsupported by Collaborator B). Collaborator B may then add a few more fillets and send the file to Collaborator A. When the incremental update is not used, the entire file is sent to Collaborator A who loses the dome as it gets converted into a loft on his/her system. This scenario is called a loop-through problem.

[0048] The present invention overcomes the loop through problem. When Collaborator B adds new fillets to the part and collaborates, the new part is compared with the one saved in the system. New fillets are detected, but the loft is not detected as changed. So only the fillets go to Collaborator A as new features and Collaborator A does not lose the dome.

[0049] In the present invention, every feature attribute is compared between the files. For example, if two extrusion features are being compared, attributes such as extrusion height, blind or through extents, and extrusion directions are compared. The profile is also checked. Each entity in the profile is compared and the coordinate system attached to the profile is also checked. If a change is found in any one of the attributes, the feature is added to the list that maintains every new, deleted and changed feature. If the base feature is changed, most (if not all) of the other features will also need to be changed. In such a situation, it is appropriate to send the entire file to the other end rather than sending a DUFO file.

[0050] The manner in which the UFO is organized is similar to the manner in which features are represented in a feature tree. This helps in the comparison, deletion and addition of features. The UFO is object oriented and extensible which aids in the easy addition of comparison operators, and comparison of disparate features with just one class. Serialization of a single feature or list of features is possible. Serialization of single feature without involving the depended features helps in making the differential update possible.

[0051] A methodology for Automated and Semi-Automated Conversion of 2D CAD Drawings to Parametric Feature-Based 3D CAD Models Via the UFO

[0052] The present invention also provides a robust and modular approach to converting drawings from 2D to 3D and consists of three primary steps in the conversion of a 2D drawing to a 3D feature based model. These steps are feature detection, feature analysis, and model creation which are further described below.

[0053] Referring now to FIG. 3 a, a flowchart 120 depicting a 2D to 3D conversion is presented. The 2D drawing 121 is processed to detect and correct any errors at step 122 . The clean drawing is then sent to an automated feature detection system at step 123 , followed by profile and feature analysis at step 124 . The result is a list of 3D features 125 that are written to the UFO at step 126 . The resultant UFO file 127 is interfaced with a commercial CAD system using an API interface, for example, as shown at step 128 . The final result is a parametric feature-based 3D model at step 129 . As part of a post-processing step, the final model is back projected to obtain the 2D drawing views at step 130 . These views are overlaid on top of the original 2D drawing views and compared at step 131 .

[0054] In a further embodiment, the method may proceed to various steps where this result is merged with a 2D representation of the UFO file to form the basis for any interactive feature definition for features that were missed by the automated conversion. In the interactive feature definition, a user could define feature profiles and interactively specify feature heights to completely define a new feature in the 2D drawing. This feature could then be added to the existing UFO file to create a new UFO file that can again be interfaced to the CAD system to generate a new 3D model.

[0055] Referring now to FIG. 3 b a system and flowchart for depicting a drawing error detection and correction 140 is presented. A 2D CAD drawing 141 is processed through an automated error detection and correction system 142 . This system 142 consists of a means to automatically detect and correct overlapping entities at step 143 , duplicate entities at step 144 , near zero length entities at step 145 and other means to recombine split entities at step 146 . The resultant 2D drawing is then processed by the semi-automated error detection and correction system 147 which consists of a means to detect disconnected entities at step 148 and crisscrossing entities at step 149 . The final result is a corrected 2D CAD drawing 150 .

[0056] Referring now to FIG. 3 c, a flowchart 160 depicting automated preprocessing steps for 2D to 3D conversions is presented. A 2D CAD drawing 161 can be formatted in various formats and is preferably a DXF or DWG format. The file 161 is read at step 162 and automated filtering of all non-graphical entities is performed at step 163 . These entities include dimension lines, centerlines, construction lines, hatching, text, title blocks and borders. Any blocks in the drawing are exploded at step 164 and error checking is performed on the preprocessed drawing at step 165 . The clean drawing is then further processed by automatically splitting the entities corresponding to the top, front and side views and a common origin is fixed for each view at step 167 . All of the entities (for example, lines, arcs and circles) are then translated to the common origin at step 168 and the data is written to classes at step 169 .

[0057] Referring now to FIG. 3 d, a system and flowchart depicting a feature detection phase 170 is presented. The first step, feature detection 172 , is the most crucial since missing features and/or wrong features can cause problems in later conversion steps. Feature detection is primarily performed as a two-step process—extraction of feature loops 173 from the 2D views and matching these loops in all views 174 . These feature loops can either be subparts (which represent joins in 3D) or nested loops (which represent cuts or holes and sometimes joins in 3D). More specifically, the feature detection means are classified into four stages—subpart extraction 173 , subpart matching 174 , nested and circular loops extraction 175 , nested and circular loops matching 176 , and corresponding feature loop matching 177 .

[0058] Referring now to FIG. 3 e, block diagrams illustrating a subpart extraction 180 is presented. Many models consist of subparts that are join features which may be a single feature or a combination of features. For example, a block 185 with a cylindrical boss or hole 186 attached would consist of two subparts. However, the block 185 could easily have additional holes within it. Those holes would be included within the block subpart. Extraction of these subparts is an important step. After initial preprocessing of an input, each 2D view consists of a nodal network with an outer boundary loop. The subpart extraction is executed for each view. This consists of a counterclockwise traversal of the outer loop's entities searching for non-trivial nodes whose directional change (the angle between two boundary entities) is less than 1800 to determine a probable subpart. After identifying the beginning point of a subpart, a traversal algorithm is executed which returns to this starting node creating the subpart loop. A subpart view is created which contains the entities of the subpart loop and any inclusive entities. This subpart extraction algorithm is repeated on every subpart until there are no more subparts being extracted for a view.

[0059] Once all subparts views have been extracted, an attempt is made to find matches for each. This is a two-step process. Initially, subparts are compared with subparts from other orthographic views looking for a possible match. Further searches look for matches using, for example, an atomic loops means. This second step is executed when there is a third view or only two views where the matching loop is nested inside the view. An example of this is the block with the cylindrical boss with three views 181 , 183 , and 185 . The cylindrical boss subpart 186 will have a subpart view (or views) associated with the rectangular extrusion loop 182 , 184 , while the remaining view will have a circle within a square. If there are three views, the two rectangular subparts will be matched during subpart matching, while the associated circle will be detected via the atomic loops means.

[0060] Once subpart extraction and matching is complete, the remaining, nested features will be detected. First, any circle loop that has not been marked as subpart match is handled. For each circle, a search is made in the orthographic views for tangential reconstruction matches, followed by tapered reconstruction matches. This will include holes, bosses, tapered holes, tapered bosses, drill tip angles, and circular fillets/chamfers.

[0061] A final step for feature detection involves a means for the nested loops. Nested loops are basically internal loops that are completely inside the outer boundary of the input views (i.e. attempts to find matches use the atomic loops). At each of the previous steps, entities are marked as ‘used’ once they have been associated with a feature (subpart or circular feature). The unused entities become the focal point of the nested loops means which eliminates unnecessary processing time and reduces the number of invalid detected features.

[0062] Nested loops are formed in a view when loops corresponding to two or more features overlap each other. The main purpose of this means is to identify the individual features in the nested loops by matching loops in available 2D views. A 3 -stage approach is followed to identify the features. The steps involved are identification of atomic loops, combination of atomic loops, and validation of combined loops. A beginning step is to identify all possible loops in each view. These loops are called atomic loops which are passed as the input for the next step. A next step is to identify the combination of loops in each view, which would correspond to a valid profile. The view in which the number of atomic loops is a minimum is taken as a reference. Each atomic loop is checked with the other view for any match. If a match was found, then the dimension of the loops is verified. If no match was found then it is considered an invalid loop. For example, if top and front view loops were to be matched then x-coordinates would be matched. If the dimension were not found to be equal, the combination of loops would be employed.

[0063] A later stage is the validation of the combined loops. In this step, the projection of the combined loop from one view is mapped on to the other view and checked for a match. The projection could be along the horizontal or vertical axis depending on the views under test. If any projection did not give a match, the loop corresponding to that projection would be removed from the combination. After the profile is validated, it is taken from one view while the height would be taken from the other view.

[0064] Upon completion of this phase, an array of matched loops which represent one or more features in different orthographic views are encountered resulting in a single matched loop that can be viewed as a back-projected representation of a 3D feature. It is not necessary to have all views represented for each feature. This allows for future expansion to additional views, in particular projected views.

[0065] Converting medium-to-complex 2D drawings with complete accuracy is difficult to achieve primarily because there may be multiple interpretations in matching feature loops in all orthographic views. Introducing human interaction at various stages of the conversion process would lead to better identification and interpretation of loop matches and would improve the percentage of drawings that can be successfully converted. As such, the present invention allows for a semi-automated approach, which provides a layer of user-interaction over the automatic software, and aids the automatic software in improving accuracy.

[0066] More specifically, the present invention introduces an application to perform 2D to 3D conversions that will allow the user control in the creation of the 3D model. This interactive process is a combination of the automated software and user controls. An initial stage consists of a feature detection step of the 2D to 3D conversion process. If the features selected are incorrect, or if one or more features are missed during the extraction process, the 3D model cannot be correctly created. A separate path is thus represented within software but is not developed as a separate application. This path can be linked in conditionally. In this manner, only one application will need to be maintained. Various tools enable the user to make selections via the desired CAD system.

[0067] The matched loops automatically detected by the software are presented one at a time to the user in the target CAD system. The user is then given the option to either accept the matched loops as such, reject them (if they are totally wrong, for example), or modify them (if minor flaws are uncovered, for example). When all of the loops are processed (either accepted, rejected or modified), the user also has the option of adding more loops, which, for example, may have been undetected by the automatic software. The conversion then proceeds to feature analysis and model build using these new matched loops.

[0068] Referring now to FIG. 3 f, a system and flowchart for the feature analysis phase 190 is presented. The array of matched loops 191 is the input to a feature analysis engine 192 . Feature analysis primarily consists of two parts. The first part, profile analysis 193 , is responsible for analyzing the actual loops of a matched loop array to look for specific feature types. Each feature, once it has completed the profile analysis stage, is turned into a feature tree such that the output of the profile analysis stage is an array of feature trees 194 . The second part, feature relationship 195 , is responsible for analyzing how each feature is related to every other feature for building the model. A description of the features is given followed by a description of each feature analysis step. 10691 The feature tree is a classic tree data structure containing a root node and children nodes where each child node, in turn, can contain more children nodes. Each node contains the necessary data to create a single 3D feature. That data consists of the matched loops, the revolve profile if necessary, a reference to the parent feature tree, the profile loop index, the associated tree level, a blind/through flag, and a pointer to the 3D feature (for example, FEAT_Base 32 ). Additionally, the tree contains two volume operation values—one indicating the relationship to the parent feature (same or opposite) and one indicating the actual operation (cut or join).

[0069] Each feature (matched loop) goes through up to three different stages in the profile analysis step. If a feature is found to be a specific type during any of the three stages, subsequent test are ignored.

[0070] The first stage is a revolve test. This means will determine if the feature is any kind of revolve feature and includes holes, bosses, tapered holes/bosses, drill tip angles, and circular fillets/chamfers as well as traditional revolve cuts and joins. Initially, the means searches for a circle loop as one of the matched loops. If there are no circle loops, the software moves on to the next profile test. However, when a circle is found, the remaining loops are tested for valid revolves. If a valid revolve is not found, one half of the loop is randomly selected to be the revolve profile, but information is propagated to the user which indicates that the revolve feature has an error in the original 2D drawing. The revolve algorithm also detects circular fillets on holes and bosses. Once complete, a tree structure is created for the feature. Child trees are created on the returned tree for circular fillets.

[0071] The second stage is a circular boss test and involves a conditional algorithm means. Inputs from the initial dialog box determine whether this algorithm is executed. The algorithm determines if a feature, while not actually a circle, can be converted to a cylindrical boss with cuts. If one loop of the matched loops contains arcs that share a center point and radius, an algorithm is executed to test if the entire loop is within a circle of the same center point and radius. This test contains a simple Boolean XOR function to determine the possible cuts, followed by a test to see if all cuts are within the circle.

[0072] If the feature is eliminated from both tests above, a third means is executed which selects one loop (based on complexity) to be the extrusion profile, and converts any other non-rectangular loops to a series of cut features. Again, a simple Boolean XOR function is executed to determine possible cuts. This means further executes an additional test to determine possible chamfers and fillets on a profile. A feature tree is created with the matched loops and their associated relationships with each other.

[0073] After the profile analysis stage is complete, an array of feature trees exist that mostly contain single feature trees with a few multi-level trees containing fillet and/or chamfer child trees. The purpose of the feature relationship software is to determine the relationship between every pair of connecting features. This is accomplished by comparing the relationship between the two features' loops containing the same coordinate system (i.e. from the same view). Loop relationships include one loop inside the second loop, one loop inside or outside the second loop with a shared edge, one loop overlapping the second loop, or two equivalent loops. In the final case, the two features are tested to possibly eliminate one feature or to combine the two features. A combination of relationships is necessary for features to be related and one relation must be overlapping or contain a shared edge. Using the previous block 185 with the cylindrical boss 186 as an example, the circle for the boss is “inside” the block's rectangular loop in one view, whereas the orthographic views contain rectangular loops for both features. The relationship between these two loops is “outside-shared” so the two features are related with the same volume operation. Had the circle loop been associated with a hole within the block instead of a boss, the rectangular loops would have had an “inside-shared” relation and the volume operation would be opposite. Once a relationship is found between two features, one feature will be have the second feature listed as its ancestor, while the second feature will have the first feature listed as its descendent.

[0074] After the first two stages are completed, all features and their relationships to one another are available to begin building the 3D model. An initial step in the model creation process is to create a single tree (if possible) from the array of trees and their relationships. The root node, or root part, of the model is first determined. A feature with no ancestors is selected as the root node feature. Once the first feature is selected, the features are added to the tree via the descendent route until no more descendents can be added. If there are features remaining that have not been added to the model tree, an ancestor feature will be added, followed by a test for the inclusion of further descendent trees. This process, one ancestor followed by descendents, will continue until all possible features have been added. As each feature is added to the model tree, the first of the volume operators is set indicating whether the child (descendent) feature is the same or an opposite operation as its parent feature. In the event that a feature is missed, that feature will be added to the error file.

[0075] Once a single 3D model tree exists, it is necessary traverse the tree again setting the final volume operation flag to cut or join. A tree function (for example, a CascadeSet function) exists which traverses through the nodes in a top down manner, setting the volume operation to cut or join depending on the parent feature and the relationship operation.

[0076] Once the 3D model is complete with the volume operations, a tree function (for example, a PopulateFeatModel function) will create a 3D model of type FEAT _— 3D_Model, for example, by converting each node of the tree to a 3D feature and adding it to the 3D model. Two functions exist, one that will create join features and a second that will create cut features. In both cases, analysis of the profile loop will determine the feature type. If the profile loop is a circle and the operation is a cut, a feature of type FEAT_Hole, for example, will be created. Before a hole is created, a test is made that checks any possible hole to see if it is a tapped hole. If the hole's profile is made up of dashed entities, and the node contains a child node with a volume operation of opposite and a profile type of circle, a test is made to see if the boundaries of the outer circle's reconstruction loops are within the inner circle's bounds. If so, a tapped hole is created combining two nodes and the tree is updated accordingly.

[0077] After a 3D model is created, a final step is executed which will combine 3D features. Recombination of features includes, two concentric connecting holes converted to counter-bore or countersink holes or a hole and a tapered hole being recombined to a hole with a drill tip angle.

[0078] Further information relating to the various features and embodiments of the present invention is available in Appendices A-D. More specifically, Appendix A describes an object-oriented analysis of three primary libraries (Geometry, Features, Geometry Network) in order to produce more robust classes. The data (attributes) associated with each of the classes, the relationships between the classes, and the classes that execute certain actions are determined. Appendix B describes multiple CAD-based applications including 2½-D editing and 3D-to-3D CAD exchanges. Since similar algorithms exist across multiple applications, libraries of related classes, their attributes and the methods (operations) that are performed on these classes are described. Appendix C describes an application to perform 2D to 3D conversions that will allow a user control in the creation of the 3D model.

[0079] Although an exemplary embodiment of the present invention has been illustrated in the accompanied drawings and described in the foregoing detailed description, it will be understood that the invention is not limited to the embodiments disclosed, but is capable of numerous rearrangements, modifications, and substitutions without departing from the spirit of the invention as set forth and defined by the following claims.