RU2450355C1 - Method of closing object contours on mesh halftone matrix - Google Patents

Abstract

FIELD: information technology.

SUBSTANCE: in the method, the size of a structure-forming element is selected equal to one element of a mesh halftone matrix; that size is increased at end points of fragments of the initial contour until the region of the structure-forming element covers the end point of the closest fragment of the initial contour, after which end points are connected by a straight line; the size of the structure-forming element is once more set equal to one element of the mesh halftone matrix and moved to the end point of the next fragment of the initial contour.

EFFECT: eliminating contour abruption on a mesh halftone matrix.

2 dwg

Description

The invention relates to systems of technical vision and can be used as a process of preliminary processing of the image to eliminate gaps in the contours of objects on the matrices of grayscale raster images of cameras when recognizing objects.

Actual bitmap images obtained from CCD video camera matrices may contain shaded and illuminated areas. Therefore, when contouring objects in contours, gaps of various sizes can occur.

There is a method of eliminating the gaps in the contours of the image of objects, based on the method of dilatation, which includes the sequential movement of a predetermined structure-forming element along the initial contour by sliding this element along the boundary of the contour of the object and closing the discontinuities of the contour of the object by overlapping them with bridges. Moreover, as a structure-forming element, you can use a circle of a size comparable to the size of the gaps (Gonzalez, R. Woods. Digital Image Processing. / Translation from English, edited by P.A. Chochia. - M .: Technosphere, 2005. - P.753 )

The main disadvantage of the method for eliminating contour discontinuities based on the dilatation operation is insufficient accuracy, since it leads to the extension of the boundaries of the object, as a result of which the contour is closed not along the initial line, but along the contour of the thickened border, which is not true.

The closest in technical essence to the claimed method (prototype) is a method of closing the contours of the image of objects, based on the combination of successive application of dilatation and erosion techniques to the image, including the sequential movement of a predetermined structure-forming element along the initial circuit. This closure method eliminates the main disadvantage of the dilatation method described above. It tends to smooth out contour sections as applied to the halftone raster image matrix, while eliminating small gaps. Structural element using arbitrary size and shape. The sequential movement of a predetermined structure-forming element along the original image contour consists in the initial sliding of the structure-forming element to eliminate discontinuities in the image contour, using dilatation, while expanding the boundaries of the contour and the subsequent sliding of the same structure-forming element in the same or reverse direction, using erosion, to narrow boundaries of the image contour, which allows you to return the position of the boundaries of the contour to its original place. As a result of applying the closure method, a smoothed closed contour of the image of the object is created (Gonzalez, R. Woods. Digital Image Processing. / Translation from English edited by P.A. Chochia. - M .: Technosphere, 2005. - S.759-760).

However, the speed of this process of closing the contours of the image of objects is insufficient due to the length of the processing time, due, firstly, to the use of resource-intensive methods of dilatation and erosion operations, which require calculations across the entire area of the structure-forming element at each point of the contour, and secondly, the need selection of the size of the structure-forming element by repeated repetition of the closure process, while striving to preserve small details, since there is a gap whose size olshe than the size selected in advance and used when performing a structure-element method is not eliminated, and all the small details in the image circuit, whose size does not exceed the size of the structure-forming element, is lost due to the smoothing circuit as a result of the closure process.

The basis of the invention is the task of improving the speed of the process of closing the contours of the image of objects, firstly, by making it possible to eliminate gaps in arbitrary sizes of the contours of the image of objects with minimal processing time and, secondly, by ensuring the safety of detail and the location of the original fragments of the image contours of objects with minimal processing time without the need for repeating circuit closure with different sizes of the structure-forming element that one.

The solution to this problem is achieved by the fact that in the method of closing the contours on the halftone raster image matrix by sequentially moving a predetermined structure-forming element along the initial contour, according to the invention, the size of the structure-forming element is chosen equal to one element of the halftone raster image matrix, this size is increased at the end points of the fragments of the initial contour until the region of the structure-forming element captures the endpoint of a nearby fragment initial contour-coagulant, whereupon these end points interconnected by a straight line. Again, the size of the structure-forming element is set equal to one element of the matrix of the halftone bitmap image and is moved to the end point of the next fragment of the original contour, continuing the process until all fragments of the original contour are processed.

Elimination of gaps in arbitrary sizes of the image contours of objects is due to the fact that the size of the structure-forming element is increased to the desired value during the closure process.

Preservation of detail and the location of the initial fragments of the image contours of objects is due to the fact that the size of the structure-forming element when moving along continuous sections of the contour is minimal and equal to one image element on a grayscale raster image matrix, and consistent application of resource-intensive methods of dilatation and erosion is not required.

The invention is illustrated by drawings, where Fig. 1 shows an example of an initial broken image contour, and in Fig. 2, the result of closing this contour by the proposed method.

A method for locking the contours of objects on a halftone raster image matrix is to select the size of the structure-forming element equal to one element of the matrix of the halftone raster image, sequentially move the structure-forming element along the initial contour, increase the size of the structure-forming element until the region of the latter captures the end point of a nearby fragment of the original contour, connecting these endpoints with each other by a straight line, followed by setting the size of the structure-forming element equal to one matrix element of the halftone bitmap image and moving to the end point of the next fragment of the original contour, continuing the process until all fragments of the original contour are processed.

The method of closing the contours of objects on the matrix of a grayscale raster image is as follows.

Initially, an arbitrary point is selected on the grayscale raster image matrix on the set of torn fragments of the original contour (Fig. 1). Since the initial arbitrary point may turn out to be non-terminal, we assume that the movement of the structure-forming element is performed clockwise. The size of the structure-forming element is set equal to one matrix element of the grayscale bitmap image, that is, to one pixel.

Carry out a sequential movement of a predetermined structure-forming element, which, in contrast to the prototype, is initially degenerate to one point along the initial contour. In the process of such movement, all the passed points are noted, continuing the process until the structure-forming element reaches the end point. If the width of the contour gap is one image element, the gap is eliminated by moving the structure-forming element to a point relative to the current one, in which it will simultaneously touch the end points of the already passed and the next contour fragment. Then this element is moved to the end point of the next contour fragment and the closure process is continued. Thus, while the contour breaks do not exceed the width of one image element, the line of the closed contour exactly repeats the lines of the original fragments, preserving all their details. In this case, the closing time is minimal, since for closing small gaps and moving along the contour, it is sufficient at each step to test on the contour image no more than seven points adjacent to the current one.

If the structure-forming element is moved to a gap larger than one image element, the coordinate value of the end point of the beginning of such a gap is stored. Then, from this point of the fragment of the initial contour, a location is made, that is, a search is made by increasing the size of the structure-forming element until the endpoint of the next regular fragment of the initial contour is hit in its region. The shape of the structure-forming element increasing in size is generally set arbitrary, in particular, having the form of a sector with a center at the end point, the central angle of which is determined by the assumed shape of the contour. If there is no information about the shape of the contour, the structure-forming element increasing in size may take the form of a circle centered at the end point. After the endpoint of another contour fragment enters the location region, the endpoint of the initial contour fragment is connected with the found endpoint of the next next fragment of the initial contour by a straight line.

Next, the structure-forming element is transferred to the found endpoint of the next fragment of the original image contour, its size is again set to one pixel and the operation of moving along the contour continues, starting from the found endpoint of the next fragment. In this way, all detected gaps on each fragment of the initial object contour on the halftone raster image matrix are eliminated. The process is continued until there is not a single untreated endpoint left to which the closure would not be applied and until the structure-forming element returns to the starting point. Thus, the original circuit is closed in all places of the detected gaps (figure 2).

Using the proposed method allows us to ensure high-quality and effective elimination of contour gaps on the halftone raster image matrix with the least complexity and without loss of detail on existing fragments of the source object's contour, which allows the most accurate recognition of objects on the halftone raster image matrix of the camera. Obviously, as follows from the above description, such contours should be non-branching structures, correspond to convex objects and have a line thickness of one pixel.

Claims (1)

A method for locking contours on a halftone raster image matrix by sequentially moving a predetermined structure-forming element along the original contour, characterized in that the size of the structure-forming element is chosen equal to one element of the halftone raster image matrix, this size is increased at the end points of the fragments of the initial contour until the region structure-forming element does not capture the endpoint of a nearby fragment of the original contour, after which these end e points are connected by a straight line, again the size of the structure-forming element is set equal to one element of the matrix of a halftone bitmap image and moved to the end point of the next fragment of the original path, continuing the process until all fragments of the original path are processed.

Images