Image processing device, edge detection method, and program

Abstract

To shorten a processing time of hysteresis threshold processing in edge detection.SOLUTION: An image processing apparatus gives, by using a first pixel in which an evaluation value of the strength of an edge is between a first threshold and a second threshold larger than the first threshold, and a second pixel in which the evaluation value is larger than the second threshold, the same label as a label indicating the second pixel or the first pixel if the label is given to any of scanned peripheral pixels p2 to p5 adjacent to a noticed pixel p1, and gives, if no label is given, a new serial symbol as a label to the noticed pixel p1, when the noticed pixel p1 is the first pixel. The image processing apparatus generates a correspondence table indicating combinations of labels in adjacent relation for each pixel to which the label is given, and determines whether or not each of the first pixels is to be an edge pixel based on the generated correspondence table.SELECTED DRAWING: Figure 10

Claims

1. An image processing apparatus that determines as edge pixels a first pixel whose evaluation value indicating edge strength falls between a first threshold and a second threshold greater than the first threshold, and a first pixel directly adjacent to a second pixel whose evaluation value is greater than the second threshold, and a first pixel indirectly adjacent to the second pixel via other first pixels, If the pixel of interest is the first pixel, and a label indicating the second pixel is assigned to any of the scanned pixels adjacent to the pixel of interest, the same label as that label is assigned to the pixel of interest; if none of the scanned pixels adjacent to the pixel of interest are assigned a label indicating the second pixel, and any of the scanned pixels adjacent to the pixel of interest are assigned a label indicating the first pixel, the same label as that label is assigned to the pixel of interest; and if none of the scanned pixels adjacent to the pixel of interest are assigned a label indicating the first pixel, a new serial number is assigned to the pixel of interest as the label indicating the first pixel. A generation means for generating a correspondence table showing the combination of labels that are adjacent to each pixel to which a label has been assigned by the aforementioned assignment means, A determination means that determines whether or not each of the first pixels is an edge pixel based on the correspondence table generated by the generation means, A preprocessing means that performs noise reduction on image data, contour extraction to extract contours from the noise-reduced image data, and non-maximal suppression to exclude areas other than those with maximum edge strength from the contour-extracted image data, Equipped with, The labeling means labels the pixels of the non-maximally suppressed image data. An image processing apparatus characterized by the following:

2. The labeling means assigns the label with the lower serial number to the pixel of interest if different labels have been assigned to each of the scanned pixels adjacent to the pixel of interest as labels indicating the first pixel. The image processing apparatus according to feature 1.

3. The labeling means, when the pixel of interest is the first pixel, assigns the same label to the pixel of interest as a label indicating an edge pixel if a label indicating the second pixel has been assigned to any of the scanned pixels adjacent to the pixel of interest. The image processing apparatus according to feature 1.

4. An edge detection method performed by an image processing device, wherein the evaluation value indicating the strength of an edge falls between a first threshold and a second threshold greater than the first threshold, and the first pixel directly adjacent to the second pixel whose evaluation value is greater than the second threshold, and the first pixel indirectly adjacent to the second pixel via other first pixels, are determined to be edge pixels. If the pixel of interest is the first pixel, and a label indicating the second pixel is assigned to any of the scanned pixels adjacent to the pixel of interest, then the same label as that label is assigned to the pixel of interest; if none of the scanned pixels adjacent to the pixel of interest are assigned a label indicating the second pixel, and any of the scanned pixels adjacent to the pixel of interest are assigned a label indicating the first pixel, then the same label as that label is assigned to the pixel of interest; and if neither a label indicating the first pixel nor a label indicating the second pixel is assigned to any of the scanned pixels adjacent to the pixel of interest, then a new serial number is assigned to the pixel of interest as a label indicating the first pixel. A generation step generates a correspondence table showing the combinations of labels that are adjacent to each pixel to which a label has been assigned in the above assignment step, A determination step is to determine whether or not each of the first pixels is an edge pixel based on the correspondence table generated in the generation step, A preprocessing step involves denoising the image data, extracting contours from the denoised image data, and then performing non-maximal suppression on the contour-extracted image data to exclude areas other than those with maximum edge strength. Includes, The aforementioned labeling step involves labeling the pixels of the non-maximal suppressed image data. An edge detection method characterized by the following features.

5. A computer for an image processing device that determines, as edge pixels, a first pixel directly adjacent to a second pixel whose evaluation value indicating edge strength is greater than the second threshold, and a first pixel indirectly adjacent to the second pixel via other first pixels, among first pixels whose evaluation value indicating edge strength is between a first threshold and a second threshold greater than the first threshold. If the pixel of interest is the first pixel, and a label indicating the second pixel is assigned to any of the scanned pixels adjacent to the pixel of interest, the same label as that label is assigned to the pixel of interest; if none of the scanned pixels adjacent to the pixel of interest are assigned a label indicating the second pixel, and any of the scanned pixels adjacent to the pixel of interest are assigned a label indicating the first pixel, the same label as that label is assigned to the pixel of interest; and if none of the scanned pixels adjacent to the pixel of interest are assigned a label indicating the first pixel or a label indicating the second pixel, a new serial number is assigned to the pixel of interest as the label indicating the first pixel. A generation means that generates a correspondence table showing the combination of labels that are adjacent to each pixel to which a label has been assigned by the aforementioned assignment means. A determination means that determines whether or not each of the first pixels is an edge pixel based on the correspondence table generated by the generation means. A preprocessing means that performs noise reduction on image data, contour extraction to extract contours from the noise-reduced image data, and non-maximal suppression to exclude areas other than those with maximum edge strength from the contour-extracted image data. To make it function as, The labeling means labels the pixels of the non-maximally suppressed image data. A program characterized by the following features.

Images