1. A static random access memory (SRAM) cell structure, comprising:
a substrate having a device isolation structure therein to define an active region, wherein the active region has a first opening and a portion of the opening is located within the device isolation structure;
a lower electrode set up within the opening;
an upper electrode set up over and completely filled the opening;
a capacitor dielectric layer set up between the upper electrode and the lower electrode; and
a transistor set up over the active region of the substrate, wherein a source region of the transistor connects with the lower electrode.
2. The SRAM cell structure of claim 1, wherein material constituting the lower electrode comprises silicon.
3. The SRAM cell structure of claim 1, wherein the capacitor dielectric layer comprises a composite silicon oxidesilicon nitridesilicon oxide stack layer.
The claims below are in addition to those above.
All refrences to claim(s) which appear below refer to the numbering after this setence.
1. A structure detection apparatus for detecting a predetermined structure in image data, the apparatus comprising:
a candidate point extraction means that extracts a plurality of candidate points belonging to the predetermined structure from the image data;
a set form storage means that stores, in advance, a set form composed of a plurality of training labels, the set form representing a known form of the predetermined structure;
a representative point selection means that selects, from the plurality of candidate points extracted by the candidate point extraction means, a plurality of representative points corresponding to the plurality of training labels respectively, the plurality of representative points composing a form model that is the same as or most similar to the set form stored in the set form storage means; and
a structure detection means that detects the predetermined structure in the image data by using the form model composed of the plurality of representative points selected by the representative point selection means, and the plurality of candidate points;
wherein the set form represents the position of each of the plurality of training labels and information about a positional relationship between the plurality of training labels, and wherein the representative point selection means selects the plurality of representative points by calculating, with respect to each combination of each of the training labels and the candidate points associated with the training labels, the degrees of correspondence between the positions of the training labels and the positions of the candidate points associated with the training labels respectively and the degree of correspondence of a positional relationship between the training labels and a positional relationship between the candidate points associated with the training labels respectively, and by determining a combination of each of the training labels and the candidate point that has the highest degree of correspondence between the positions thereof and the highest degree of correspondence between the positional relationships.
2. A structure detection apparatus, as defined in claim 1, wherein the candidate point extraction means detects a region having an image characteristic of the predetermined structure in the image data without using the set form, and extracts the plurality of candidate points from the detected region.
3. A structure detection apparatus, as defined in claim 1, wherein the set form further represents information about a connection relationship between the training labels, and wherein the representative point selection means determines, for each of the combinations, a path between two of the candidate points associated with two of the training labels that should be connected to each other in such a manner to selectively trace a plurality of candidate points including a candidate point that is not associated with the training label in the combinations so that a cost based on a predetermined index value becomes lowest, and calculates the degree of correspondence of the positional relationships further by taking the cost for the path into consideration, and wherein the structure detection means detects the predetermined structure based on a manner of tracing the candidate points between the representative points in the combinations determined by the representative point selection means.
4. A structure detection apparatus, as defined in claim 1, wherein the structure detection means detects the predetermined structure by correcting a form that connects the plurality of candidate points by using the form model that connects the representative points in such a manner to form substantially the same form as the set form.
5. A structure detection apparatus, as defined in claim 4, wherein the structure detection means connects the plurality of candidate points by using Spanning Tree algorithm.
6. A structure detection apparatus, as defined in claim 1, wherein when a defect region that has been detected as the predetermined structure in the form model connecting the representative points, but in which the candidate points discontinue, is present, the structure detection means detects the defect region as the predetermined structure.
7. A structure detection apparatus, as defined in claim 1, wherein when an excess region in which the candidate point has been detected, but which has not been detected as the predetermined structure in the form model, is present, the structure detection means excludes the excess region from the predetermined structure.
8. A structure detection apparatus, as defined in claim 1, wherein the representative point selection means selects, by graph matching, a set of representative points composing the form model that is the same as or most similar to the set form.
9. A structure detection apparatus, as defined in claim 1, wherein the set form storage means stores an evaluation function representing a likelihood that the candidate points form the set form, the evaluation function having been learned by using training image data that have been known to represent the predetermined structure, and wherein the representative point selection means selects the representative points by using the evaluation function.
10. A structure detection apparatus, as defined in claim 1, wherein when the predetermined structure is a blood vessel, the representative point selection means selects the representative points by using the thickness of a blood vessel and a luminance at each of the candidate points together with the coordinate of each of the candidate points.
11. A structure detection apparatus, as defined in claim 1, the apparatus further comprising:
a normalization means that normalizes, based on a predetermined reference position, the coordinates of the plurality of candidate points extracted by the candidate point extraction means.
12. A structure detection apparatus, as defined in claim 1, wherein the candidate point extraction means extracts a plurality of points from a region of the predetermined structure, and connects the extracted plurality of points to each other, and divides the extracted plurality of points into segments each having a predetermined length, and extracts the candidate points from the divided segments respectively.
13. A structure detection apparatus, as defined in claim 1, wherein the set form is composed of a plurality of structures including the predetermined structure, the apparatus further comprising:
a display control means that displays structures detected by the structure detection means in such a manner that different structures are distinguishable from each other.
14. A structure detection apparatus, as defined in claim 1, wherein the set form includes not only the form of the predetermined structure but also the form of a structure that is different from the predetermined structure, and wherein the candidate point extraction means extracts the plurality of candidate points from the predetermined structure and the different structure, and wherein the representative point selection means selects the plurality of representative points composing the form model that is the same as or most similar to the set form, and wherein the structure detection means deletes the representative points corresponding to the different structure from the plurality of representative points, and detects the predetermined structure.
15. A structure detection apparatus, as defined in claim 1, wherein the predetermined structure is a tubular structure in a human body.
16. A structure detection method for detecting a predetermined structure in image data, the method comprising the steps of:
extracting a plurality of candidate points belonging to the predetermined structure from the image data;
selecting, from the extracted plurality of candidate points, a plurality of representative points corresponding to a plurality of training labels respectively, the plurality of representative points composing a form model that is the same as or most similar to a set form that is composed of the plurality of training labels, the set form representing a known form of the predetermined structure and having been stored in advance in a set form storage means; and
detecting the predetermined structure in the image data by using the form model composed of the selected plurality of representative points, and the plurality of candidate points;
wherein the set form represents the position of each of the plurality of training labels and information about a positional relationship between the plurality of training labels, and wherein the plurality of representative points are selected by calculating, with respect to each combination of each of the training labels and the candidate points associated with the training labels, the degrees of correspondence between the positions of the training labels and the positions of the candidate points associated with the training labels respectively and the degree of correspondence of a positional relationship between the training labels and a positional relationship between the candidate points associated with the training labels respectively, and by determining a combination of each of the training labels and the candidate point that has the highest degree of correspondence between the positions thereof and the highest degree of correspondence between the positional relationships.
17. A non-transitory computer-readable recording medium storing therein a structure detection program for causing a computer to execute processing for detecting a predetermined structure in image data, the program comprising the procedures of:
extracting a plurality of candidate points belonging to the predetermined structure from the image data;
selecting, from the extracted plurality of candidate points, a plurality of representative points corresponding to a plurality of training labels respectively, the plurality of representative points composing a form model that is the same as or most similar to a set form that is composed of the plurality of training labels, the set form representing a known form of the predetermined structure and having been stored in advance in a set form storage means; and
detecting the predetermined structure in the image data by using the form model composed of the selected plurality of representative points, and the plurality of candidate points;
wherein the set form represents the position of each of the plurality of training labels and information about a positional relationship between the plurality of training labels, and wherein the plurality of representative points are selected by calculating, with respect to each combination of each of the training labels and the candidate points associated with the training labels, the degrees of correspondence between the positions of the training labels and the positions of the candidate points associated with the training labels respectively and the degree of correspondence of a positional relationship between the training labels and a positional relationship between the candidate points associated with the training labels respectively, and by determining a combination of each of the training labels and the candidate point that has the highest degree of correspondence between the positions thereof and the highest degree of correspondence between the positional relationships.