Inspection apparatus

1. An inspection apparatus for performing an appearance inspection using a plurality of images obtained by imaging an inspection target, the inspection apparatus comprising:

a determination unit that

estimates an inspection result on each of the plurality of images based on a predetermined relationship model between the image and the inspection result, and

calculates a comprehensive inspection result based on reliability of each of the inspection results on the plurality of images,

wherein

when the reliability of the comprehensive inspection result does not satisfy a certain condition, the determination unit

outputs an instruction to re-image the image, and

calculates the comprehensive inspection result again using the re-imaged image,

the inspection apparatus further comprises an imaging condition decision unit that determines an imaging condition of the image,

the imaging condition decision unit changes an imaging condition at the time of the re-imaging from an imaging condition at the time of previous imaging, and

the imaging condition decision unit increases the number of images to be taken at the time of the re-imaging from the number of images to be taken at the time of previous imaging for an inspection part where the reliability of the inspection result does not satisfy a certain condition.

2. An inspection apparatus for performing an appearance inspection using a plurality of images obtained by imaging an inspection target, the inspection apparatus comprising:

a determination unit that

estimates an inspection result on each of the plurality of images based on a predetermined relationship model between the image and the inspection result, and

calculates a comprehensive inspection result based on reliability of each of the inspection results on the plurality of images,

wherein

when the reliability of the comprehensive inspection result does not satisfy a certain condition, the determination unit

outputs an instruction to re-image the image, and

calculates the comprehensive inspection result again by using the re-imaged image and the image imaged up to the previous time,

the inspection apparatus further comprises an imaging condition decision unit that determines an imaging condition of the image,

the imaging condition decision unit changes an imaging condition at the time of the re-imaging from an imaging condition at the time of previous imaging, and

the imaging condition decision unit increases the number of images to be taken at the time of the re-imaging from the number of images to be taken at the time of previous imaging for an inspection part where the reliability of the inspection result does not satisfy a certain condition.

3. An inspection apparatus for performing an appearance inspection using a plurality of images obtained by imaging an inspection target with different imaging conditions, the inspection apparatus comprising:

a determination unit configured to

estimate an inspection result on each of the plurality of images imaged with the different imaging conditions based on a learning model which is previously constructed by learning a correlation between (i) feature quantity based on data of the image and (ii) the inspection result representing a result of quality determination, and

calculate a comprehensive inspection result based on reliability relating to a “No Good” or “Good” degree of evaluation for each of the estimated inspection results on the plurality of images imaged with the different imaging conditions, wherein

the reliability relating to the “No Good” or “Good” degree of evaluation is an index indicating the likelihood of failure with numerical value,

wherein the inspection apparatus further comprises an imaging condition decision unit configured to determine and change an imaging condition of the images to be taken at the time of the re-imaging from the number of images to be taken at the time of previous imaging for an inspection part where the reliability of the inspection result does not satisfy a certain condition, according to the reliability relating to the “No Good” or “Good” degree of evaluation, and

wherein the inspection apparatus is configured to re-image and re-evaluate the comprehensive inspection result until the comprehensive inspection result with a predetermined reliability is obtained.

4. The inspection apparatus according to claim 3 , wherein

in response to the reliability of the calculated comprehensive inspection result not satisfying a certain condition, the determination unit is configured to

output an instruction to re-image the image, and

calculate the comprehensive inspection result again using the re-imaged image.

5. The inspection apparatus according to claim 4 , wherein the imaging condition decision unit is configured to increase the number of images to be taken at the time of the re-imaging from the number of images to be taken at the time of previous imaging for an inspection part where the inspection result is not good.

6. The inspection apparatus according to claim 4 , wherein

the determination unit has a plurality of different learning models, the plurality of different learning models including the previously constructed learning model, and

the determination unit is configured to estimate the inspection result for the image at the time of the re-imaging, using a learning model different from the previously constructed learning model based on which the inspection result was estimated for the image at the time of previous imaging.

7. The inspection apparatus according to claim 3 , wherein

in response to the reliability of the calculated comprehensive inspection result not satisfying a certain condition, the determination unit is configured to

output an instruction to re-image the image, and

calculate the comprehensive inspection result again by using the re-imaged image and the image imaged up to the previous time.

8. The inspection apparatus according to claim 7 , wherein the imaging condition decision unit is configured to increase the number of images to be taken at the time of the re-imaging from the number of images to be taken at the time of previous imaging for an inspection part where the inspection result is not good.

9. The inspection apparatus according to claim 7 , wherein

the determination unit has a plurality of different learning models, the plurality of different learning models including the previously constructed learning model, and

the determination unit is configured to estimate the inspection result for the image at the time of the re-imaging, using a learning model different from the previously constructed learning model based on which the inspection result was estimated for the image at the time of previous imaging.

10. The inspection apparatus according to claim 3 , wherein

the determination unit has a plurality of different learning models, the plurality of different learning models including the previously constructed learning model, and

the determination unit is configured to estimate the inspection result using the previously constructed learning models in accordance with each of the plurality of images.

11. The inspection apparatus according to claim 3 , wherein the determination unit is further configured to classify, based on (1) the index indicating the likelihood of failure or (2) statistic amounts of a plurality of inspection results on the plurality of images imaged with the different imaging conditions, the comprehensive inspection result into “not good”, “indefinite”, and “good”.