JPH026103B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION (1) Technical Field of the Invention The present invention relates to a fish head cut position detection circuit in an apparatus for automatically processing fish bodies.

(2) Conventional technology and problems Conventionally, in order to detect the position of the decapitation of the fish body in the process of sending collected fish to an automatic processing device, the position of the gills was estimated from the positional relationship of the head or head and tail. There are methods for extracting features from binarized fish images, but the former has the disadvantage of large errors, and the latter has the disadvantage of a complicated extraction process and lack of stability.

(3) Object of the Invention The object of the present invention is to provide a fish head/tail cutting position detection circuit that evaluates and determines the optimum head cutting position from a grayscale image of the fish body.

(4) Structure of the invention In order to achieve the above object, the fish head cut position detection circuit of the present invention includes a head-to-tail detection circuit that detects the head and tail of the fish from image data, and a head-to-tail detection circuit that detects the head and tail of the fish from image data; an average concentration distribution detection circuit that detects a one-dimensional average concentration distribution along the cranial-caudal direction of the fish body by scanning in the direction; a noise removal circuit that removes noise from the one-dimensional data output from the average concentration distribution circuit; An edge detection circuit detects edges of the one-dimensional data from which noise has been removed and uses them as candidates for cutting, and compares the peak value with a threshold, and if the edge is larger than the threshold, An evaluation/judgment circuit that determines the candidate as a cutting position, and if it is smaller, determines the cutting position by a formula using the head position, tail position, and constants set in advance for the type of fish body. It is something to do.

(5) Embodiment of the invention FIG. 1 is an explanatory diagram of the configuration of an embodiment of the invention.
First, after obtaining two-dimensional contour shading data by optically scanning the fish body using a preprocessing circuit (not shown),
The dimensional image data is input to the average density distribution detection circuit 1, and the density difference representing the fish body is averaged along a line perpendicular to the side line of the fish body outline, and is projected onto a straight line parallel to the side line. Find the average concentration distribution of
This average concentration distribution data is input to a noise removal circuit 2, and random noise is removed and smoothed using, for example, a median filter. It is known that the concentration at the position where the gills normally exist varies greatly compared to other parts of the fish body, and based on this, the concentration can be obtained by scanning in the row direction of the fish body, that is, in the dorsoventral direction. A first-order edge detection circuit extracts a portion of the average density distribution data as a candidate point for cutting by using a primary edge detection circuit, and the peak value of the candidate point is measured and defined as "certainty."

Note that the higher the peak value of this certainty, the more effective the certainty becomes.

On the other hand, the two-dimensional image data is branched and input to the head/tail detection circuit 4, and the head position and tail position are detected from the outline data of the fish body. Any method may be used to detect the head position and the tail position, but in the present invention, as an example, the patent application No. 1982-1 filed by the present applicant on June 18, 1982 is used.
The technology related to head and tail detection in No. 104745 (Fish direction identification device) is adopted.

Input the head position and tail position from the head detection circuit 4 into the evaluation circuit 5-1 of the evaluation and judgment circuit 5, and input candidate points and certainty information from the edge detection circuit 3.
Based on these, using a predetermined evaluation function or determination function, the determination circuit 5-2 determines and outputs the position of head amputation.

FIG. 2 shows a specific circuit example of the evaluation and determination circuit 5. If the candidate CP' of the head cutting position from the edge detection circuit 3 in FIG. Predetermined threshold Th
If it is larger, that is, |S|≧Th, the detection result CP of the cutting position is CP=CP′ (1).

In addition, when |S|<Th, CP=(TP−HP)×N+HP (2) Here, N is a multiplier set in advance for each type of fish.

By using the above equations (1) and (2), it is possible to ensure an almost accurate head cutting position.

The figure uses equations (1) and (2), puts the probability S and its threshold value Th into the comparison circuit 11, checks the magnitude of |S| and Th, and uses equation (1) when |S|≧Th. candidate by
CP' is output as the detected cutting position CP, and |S|<Th
In this case, the head position (HP) and tail position (TP) are calculated by the calculation circuit 12 according to equation (2), and (TP
-HP)×N+HP is output as the detected cutting position CP.

(6) Effects of the Invention As explained above, according to the present invention, the average density distribution is obtained from the grayscale image of the fish body, and the candidate points of the cutting position, the probability, and the head-to-tail position obtained by edge detection are calculated from the density image of the fish body. Based on the position, the optimal head cutting position can be detected according to predetermined evaluation criteria. As a result, if the probability of a candidate point is more than a threshold and it is certain, it is cut at that candidate point, and if the probability is less than a threshold and uncertain, an estimation method from the head-to-tail positional relationship is used. It is comprehensively reliable and can simplify the extraction process.

[Brief explanation of drawings]

FIG. 1 is a configuration explanatory diagram of an embodiment of the present invention, and FIG. 2 is a specific circuit example of the main part of the embodiment of FIG. 1. In the figure, 1 is an average concentration distribution detection circuit, 2 is a noise removal circuit, 3 is an edge detection circuit, 4 is a head/tail detection circuit, 5
11 is a comparison circuit, 12 is an arithmetic circuit, and 13 is a selection circuit.

Claims (1)

[Scope of Claims] 1. A head-to-tail detection circuit that detects the head and tail of a fish body from image data, and a one-dimensional average density distribution along the cranial and caudal direction of the fish body by scanning the image data in the dorsal and ventral direction of the fish body. an average density distribution detection circuit that detects the average density distribution; a noise removal circuit that removes noise from the one-dimensional data output from the average density distribution circuit; and a noise removal circuit that detects edges of the noise-removed one-dimensional data and uses them as candidates for cutting. At the same time, an edge detection circuit that measures the peak value of the edge compares the peak value with a threshold value, and if the value is larger than the threshold value, the candidate is determined to be the cutting position, and if smaller, the candidate is determined to be the cutting position. A head-and-tail cut position detection circuit for a fish body, comprising: an evaluation/judgment circuit for determining a cut position using a formula using a constant set in advance for each type of fish body;