According to one embodiment a necking machine is provided for containers with an open end. The necking machine may comprise a plurality of holding stations for containers and a plurality of tool stations contiguous and opposing the holding station. The tool stations being movable with respect to each other. The machine also includes an inspection device for containers that includes a light detector for detecting light in the interior of each container. The inspection device is attached to a tool station such that the inspection device shifts together with the tool station with respect to the container during the inspection of the container. Light communication of the light detector with the interior of the container being maintained during the inspection.

According to one embodiment a necking machine is provided for containers with an open end. The necking machine may comprise a plurality of holding stations for containers and a plurality of tool stations contiguous and opposing the holding station. The tool stations being movable with respect to each other. The machine also includes an inspection device for containers that includes a light detector for detecting light in the interior of each container. The inspection device is attached to a tool station such that the inspection device shifts together with the tool station with respect to the container during the inspection of the container. Light communication of the light detector with the interior of the container being maintained during the inspection.

A defect classification method in accordance with the present invention uses two types of images output from the defect inspection device 150 (i.e., the first inspection image generated from a luminance signal sequentially output from a detector SE and the second inspection image generated from a difference of the signals from an adjacent portion in a region where the defect exists). The first inspection image includes information for discriminating unevenness of the defective shape. Also, while it is difficult to discriminate unevenness of the defective shape by the second inspection image, the second inspection image includes information on a luminance distribution emphasizing a defective section. The region of the defective section is extracted from the second inspection image to be applied to the first inspection image and thereby define an arithmetic processing area, and the image processing is performed within the arithmetic processing area to compute a feature amount.

A defect classification method in accordance with the present invention uses two types of images output from the defect inspection device 150 (i.e., the first inspection image generated from a luminance signal sequentially output from a detector SE and the second inspection image generated from a difference of the signals from an adjacent portion in a region where the defect exists). The first inspection image includes information for discriminating unevenness of the defective shape. Also, while it is difficult to discriminate unevenness of the defective shape by the second inspection image, the second inspection image includes information on a luminance distribution emphasizing a defective section. The region of the defective section is extracted from the second inspection image to be applied to the first inspection image and thereby define an arithmetic processing area, and the image processing is performed within the arithmetic processing area to compute a feature amount.

The methods herein utilize polarized light for detecting through holes in substrates. A light source directs light through a first polarization filter having a first polarization axis, wherein polarized light travels from the first polarization filter and toward a substrate. The orientation of the polarized light is changed while traveling through substrate material, and is scattered. However, polarized light traveling through a hole in the substrate remains unscattered. A second polarization filter receives unscattered light and scattered light traveling away from the substrate. The second polarization filter includes a second polarization axis angularly offset from and not parallel with the first polarization axis. As such, the second polarization filter blocks the advancement of unscattered light while the scattered light is not blocked by the second polarization filter. The hole is detected based on an absence of unscattered light surrounded by light traveling from the second polarization filter.

The methods herein utilize polarized light for detecting through holes in substrates. A light source directs light through a first polarization filter having a first polarization axis, wherein polarized light travels from the first polarization filter and toward a substrate. The orientation of the polarized light is changed while traveling through substrate material, and is scattered. However, polarized light traveling through a hole in the substrate remains unscattered. A second polarization filter receives unscattered light and scattered light traveling away from the substrate. The second polarization filter includes a second polarization axis angularly offset from and not parallel with the first polarization axis. As such, the second polarization filter blocks the advancement of unscattered light while the scattered light is not blocked by the second polarization filter. The hole is detected based on an absence of unscattered light surrounded by light traveling from the second polarization filter.

Provided is a sheet inspection device capable of saving space, reducing member costs, and reducing the number of maintenance steps. A first light source Ls1 and a first inspection part S1 for inspecting a flaw on the front surface of a sheet and a second light source Ls2 and a second inspection part S2 for inspecting a flaw on the back surface of the sheet are disposed in such a positional relationship that when the sheet has a hole, the first inspection part S1 can detect light emitted by the second light source and transmitted through the hole in the sheet.

A defect classification method in accordance with the present invention uses two types of images output from the defect inspection device 150 (i.e., the first inspection image generated from a luminance signal sequentially output from a detector SE and the second inspection image generated from a difference of the signals from an adjacent portion in a region where the defect exists). The first inspection image includes information for discriminating unevenness of the defective shape. Also, while it is difficult to discriminate unevenness of the defective shape by the second inspection image, the second inspection image includes information on a luminance distribution emphasizing a defective section. The region of the defective section is extracted from the second inspection image to be applied to the first inspection image and thereby define an arithmetic processing area, and the image processing is performed within the arithmetic processing area to compute a feature amount.

A method of film production includes the steps of obtaining defect information including information on the position of a defect (D) in a separator original sheet (12b), slitting the separator original sheet (12b) to produce a plurality of separators (12a), and determining, on the basis of the defect information on a single defect (D), that a separator (12a) actually having the defect (D) and another separator (12a) adjacent to the above separator (12a) are defective.

A method of film production includes the steps of obtaining defect information including information on the position of a defect (D) in a separator original sheet (12b), slitting the separator original sheet (12b) to produce a plurality of separators (12a), and determining, on the basis of the defect information on a single defect (D), that a separator (12a) actually having the defect (D) and another separator (12a) adjacent to the above separator (12a) are defective.