Introduction

Remote sensing technology is a long-range mean of earth observation. Using remote sensing technology can effectively observe a wide range of ground object information and provide a database for ocean island research. Remote sensing images have the characteristics of wide-area observation, real-time, and multi-band. By analyzing remote sensing images, it is possible to quickly obtain information on the location, boundary, and land species types of ocean islands based on remote sensing images in real time and help to detect them. On the other hand, remote sensing images have the characteristic of long-time series. By comparing and analyzing remote sensing images at different times, it is possible to search for important information such as ocean detection. Remote sensing provides an important ocean detection database, which can be used as an important mean of data collection. Based on the multi-band of remote sensing image, it can be used to improve the conventional image segmentation method. Based on the research on the ocean detection of remote sensing image boundary features, a real-time detection method covering a large area is proposed, and an ocean detection algorithm is proposed (Demontoux et al. 2007). Based on image detection, in order to meet various application requirements, high-speed ocean detection based on remote sensing image segmentation is realized (Despagne 2006). The ocean provides us with a large amount of living resources and production resources. However, due to environmental pollution, ocean tides, and other changes in the marine environment, the habitat and reproduction of marine organisms have been affected, thereby endangering people’s lives and work (Doolittle and Brevik 2014). China has jurisdiction over 18,000 kilometers of coastline and approximately 3 million square kilometers of sea area. In order to be able to conduct comprehensive inspections of the ocean, high-tech technologies must be adopted (Douaoui et al. 2006). The Chinese government mentioned “marine monitoring technology” in the Ninth Five-Year Research Plan of the National 863 Program (Fao 2003). In order to promote the development of China’s marine detection technology, the technology is the subject of the National 863 Program, “investing 120 million yuan during the research period of high-level marine detection technology (Fatás et al. 2013).” In order to strengthen the high-tech research of marine monitoring, during the “Eleventh Five-Year Plan” period, greater investment will be made to enable continuous innovation of marine environmental monitoring technology (Gallali 1980). In the ever-evolving social and economic context, foreign language translations with strong professional capabilities and high comprehensive quality are required (Gascuel-Odoux and Boivin 1994). In the translation talent market, the social needs of Japanese translation cannot be ignored. The exchanges and cooperation between China and Japan in different fields require a higher level of Japanese translation (Ghorbani et al. 2019). Therefore, it is very important to conduct in-depth discussions on the status quo of Japan’s trade and translation talents (Goovaerts 1997).

Materials and methods

Research area and data sources

Remote sensing images provide a wealth of information for the study of ocean internal waves. The sea areas surrounding China, especially area A, are areas where domestic waves frequently occur (as shown in Fig. 1). The northern internal waves of A appear not only in the form of solitary waves, but also in the form of internal tsunamis (Gorji et al. 2015). This article focuses on the internal ocean waves with central latitude and longitude of 21° N and 116° E that occurred in area A at 14:11 on August 15, 2006, and selected SAR images. It can be seen from Fig. 2 that the framed area is the area with more obvious internal waves (Goulet and Barbeau 2004).

Fig. 1
figure 1

Topographic map of the study area

Full size image
Fig. 2
figure 2

Unprocessed SAR image of internal ocean waves

Full size image

Filter preprocessing

The inherent characteristics of SAR coherent imaging mechanism make SAR images produce speckle noise. As a result, image segmentation and target detection will become more difficult, so noise suppression needs to be performed first.

The Sarscape module in ENVI software can reduce noise through multi-view processing, but this will reduce image resolution and loss of detail. The classic filtering methods such as Frost, Lee, and Kuan in the AENVI software use a fixed window to estimate the image statistical parameters and then achieve filtering (Goulet and Barbeau 2006). Therefore, it is difficult to balance between coherent speckle suppression and edge preservation.

This article uses the Lee filter of the default 3 × 3 window in the ENVI software to preserve the edge information of the image as much as possible. In the analysis, a variety of window combinations were selected, namely, 3 × 3, 5 × 5, and 7 × 7. Since different window sizes have little impact on subsequent analysis, the image noise is significantly reduced after the filtering process in Fig. 3, and the detailed information of the image is not lost, as shown in the framed area in Fig. 3 as the contour of the ocean internal waves (Gratton 2002).

Fig. 3
figure 3

Filter preprocessing results of marine internal wave SAR image

Full size image

Improved binarization Gaussian filtering level set model design and accuracy detection

By combining the advantages of the GAC model and the C-V model, the hierarchical set evolution equation of the binary Gaussian filtering model can be described as follows:

$$ \frac{\partial \phi }{\partial t}=\mathrm{sp}f\left(I(x)\right)\mid \nabla \phi \mid \alpha $$
(1)

Among them, ϕ(x) is the level set function; α is the constant velocity; spf is the symbolic pressure function. SPF can be expressed as:

$$ \mathrm{spf}\left(I(x)\right)=\frac{\mathrm{I}-\frac{c_1+{c}_2}{2}}{\max \left(\left|\mathrm{I}-\frac{c_1+{c}_2}{2}\right|\right)} $$
(2)

By combining the segmentation results of OTSU and the non-linearly optimized K-means, this paper proposes an improved initial evolution curve as the input of the level set Eq. (1) of the SGBFRLS model. The specific algorithm is as follows:

Set the initial evolution curve as C = cotsu∪ckmeans; C satisfies the following equation:

$$ \left\{\begin{array}{c}C=\left\{x\in \Omega :\phi \left(\mathrm{x}\right)=0\right\}\\ {}\mathrm{inside}(C)=\left\{x\in \Omega :\phi \left(\mathrm{x}\right)>0\right\}\\ {}\mathrm{outside}(C)=\left\{x\in \Omega :\phi \left(\mathrm{x}\right)<0\right\}\end{array}\right. $$
(3)

In order to compare the segmentation accuracy of the three models, the experimental results of the C-V model, the SBGFRLS model, and the I-SBGFRLS model were compared from the three aspects of shape, length, and area.

In order to verify the accuracy of segmentation, three accuracy verification parameters are proposed, and the parameter ranges are as follows:

$$ a=\frac{S_0}{S_1} $$
(4)
$$ {a}_1=\frac{S_0}{S_2} $$
(5)
$$ \Delta =\left|\frac{S_0}{S_1}-\frac{S_0}{S_2}\right| $$
(6)

In the area intersection, the automatic extraction area is the manual interpretation area. When the shapes are similar, the more Δ tends to 0, the higher the automatic extraction accuracy.

Results

Analysis of the basic situation of the study area

This chapter uses two remote sensing images of different sizes for experiments. The size of image one is 600 × 600; pixels: the size of image two is 808 × 1182 pixels (Fig. 4). As shown in Fig. 5, this is the composite effect of bands 4, 6, and 5.

Fig. 4
figure 4

Experimental data (a image one; b image two)

Full size image
Fig. 5
figure 5

Visual interpretation of perimeter and area diagram (a image one boundary; b image one area; c image two boundary; d real image two area)

Full size image

In Table 1, the perimeter and area of images one and two are counted, and the segmentation results under different segmentation algorithms are compared.

Table 1 Perimeter and area statistics of experimental data based on visual interpretation (unit: number of pixels)
Full size table

Analysis of ocean boundary extraction results of different algorithms

The I-SBGFRLS model proposed in this chapter is an improved version of the SBGFRLS model. The SBGFRLS model is improved by combining the advantages of the GAC model and the C-V model. Therefore, the comparison experiment selected the C-V model, the SBGFRLS model, and the I-SBGFRLS model. In order to compare the computational efficiency of different models, three models were compared in terms of time and number of repetitions.

In Fig. 6, the first line is the initial contour, and the second line is the calculation result. From left to right, there are three models: C-V, SBGFRLS, and I-SBGFRLS. According to the segmentation result of Fig. 5, the following can be seen. (1) The initial contours of the C-V model and the SGBFRLS model are far away from the target boundary. (2) In the first and second image segmentation of the C-V model, over-segmentation occurred, and a part of the island was segmented. (3) The calculation results of the SBGFRLS model and the I-SBGFRLS model are similar, and both can get good results as well as the boundary of the island.

Fig. 6
figure 6

Experimental results of different algorithms (a image one; b image two)

Full size image

From Table 2, the following things can be known. (1) In the case of processing two images with different sizes, the C-V model has the most calculation time and number of repetitions, and the calculation efficiency is low. (2) When the SBGFRLS model processes two images of different sizes, the number of repetitions will be close to each other, but the calculation time is quite different. (3) The initial expansion curve of the I-SBGFRLS model is close to the target boundary, so the calculation time and number of iterations of the I-SBGFRLS model are the least, and the calculation efficiency is the highest.

Table 2 Comparison of the efficiency of the three segmentation models
Full size table

From Table 3 and Table 4, the following can be known. (1) The segmentation accuracy of image 1 with a smaller image size of the C-V model is higher, but the segmentation accuracy of image 2 with a larger image size is greatly reduced. (2) The SBGFRLS model can achieve good segmentation results for images 1 and 2 of different sizes. (3) The I-SBGFRLS model can achieve better segmentation results for the first and second images of different sizes. Through the above experiments, the improved binarized Gaussian filter level setting model proposed in this chapter proves that the separation efficiency can be greatly improved without changing the accuracy.

Table 3 Results of automatic extraction and manual interpretation (unit: pixel)
Full size table
Table 4 Verification table of similarity between automatic extraction results and manual interpretation results (precision: 0.00)
Full size table

The detection and processing steps of remote sensing image boundary characteristics

Because SAR images are very noisy, and the main function of column separation neighborhood processing is to simplify the complex gray value of the image, this paper performs column separation neighborhood processing on marine internal wave SAR images.

The result of column separation neighborhood processing on Fig. 3 is shown in Fig. 7. It can be seen from Fig. 7 that after the column separation neighborhood processing, the processed image becomes more rounded, the edges are softer, and the pixel gray value of the entire image has undergone a fundamental change.

Fig. 7
figure 7

Array separation neighborhood processing results of marine internal wave SAR image

Full size image

In this paper, when the Canny operator edge detection is performed on the internal wave image, a variety of thresholds and variances are used for comparative analysis. Among them, 0.1 ≤ Threshold ≤ 0.6, the step size is 0.05; 10 ≤ Variance ≤ 60, the step size is 5, so there are a total of 121 combinations. Figure 8 is an image detected by the Canny operator when the threshold is 0.2 and the variance is 30.

Fig. 8
figure 8

The edge detection result of Canny operator in ocean internal wave SAR image

Full size image

It can be seen from Fig. 8 that some of the detected contours in the figure are not internal wave light and dark stripes, so further processing is still needed.

Before processing, analyze the threshold and variance parameters in the Canny algorithm. Under 121 combinations of different thresholds and variances, the changes in the maximum length and number of stripes are drawn into three-dimensional graphics (Fig. 9a, b), and the contour recognition effect under each combination is compared to find the best threshold and variance parameters. In Fig. 9c, d, along the positive horizontal axis, every 11 points are the same threshold, divided into 11 segments, that is, a total of 121 points. It can be seen from Fig. 9d that under the same threshold, as the variance gradually increases, the number of fringes gradually decreases. By comparing the Canny detection effect map under 121 combinations, it can be seen that when the extreme value of the maximum length and number of stripes tends to be stable, take the threshold and variance value corresponding to the first time, the stable value is reached, and it can be found that the recognition effect near the threshold and variance is better. Therefore, on this basis, the threshold and variance values can be fine-tuned to achieve better detection results, thereby obtaining the threshold and variance parameters in Fig. 8.

Fig. 9
figure 9

Comparison of the number and length of fringes detected under different thresholds and variances

Full size image

We analyze the contour in Fig. 10 and draw the smallest circumscribed rectangle and circumscribed circle. The result is shown in Fig. 11. According to the shape of the five types of contours and the length-to-width ratio of the minimum circumscribed rectangle, for the sake of caution, this paper takes r = 0.3.

Fig. 10
figure 10

Colored marker map of the outline

Full size image
Fig. 11
figure 11

The minimum circumscribed rectangle and circumscribed circle of the outline

Full size image

It can be seen from Fig. 10 that the vertical direction of the internal wave profile reflects the general direction of internal wave propagation. In an internal wave image, the propagation direction of the internal wave changes little, so the contour characteristics of the ocean internal wave can be screened based on this feature. Perform a straight line fitting on each contour in Fig. 10, turn the curve into a straight line, obtain the straight line equation and the vertical line equation, and then obtain the θ parameter. Analyzing the contour in Fig. 10, the result of fitting a straight line is shown in Fig. 12. For the sake of beauty, only part of the fitted straight line and its corresponding vertical line are drawn.

Fig. 12
figure 12

Fitted line and vertical line diagram of partial contours

Full size image

Taking horizontal to the right as the horizontal axis and vertical to the bottom as the vertical axis, the four fitted straight line equations in Fig. 12 are y = k(x–x_0)+y_0, where k is the slope of the fitted straight line, and x_0 and y_0 are on the straight line. The coordinates at the center of the contour.

According to the parameters of the fitted straight line equation, the direction of the vertical line in each contour in Fig. 12 can be obtained, that is, the slope of the vertical line (k = − 1/k). It can also be seen from Fig. 12 that the angle between the vertical direction of the internal wave light and dark stripes and the horizontal direction is within a certain range.

In summary, by judging and filtering the contour length, area ratio and direction, the fringe profile of the real internal wave that meets the conditions is shown in Fig. 13a, and the real internal wave fringes are extracted as shown in Fig. 13b. At the same time, according to the vertical direction parameter k of the internal wave fringe profile, the ocean internal wave packet group can be further subdivided, as shown in the upper left corner of the ocean internal wave profile in Fig. 13, where the vertical direction parameter value k? is at a very high level. Changes within a small range means that they belong to the same wave packet group. In addition, the latitude and longitude information of each fringe is extracted according to the identified real internal wave fringes (as shown in Fig. 14).

Fig. 13
figure 13

Recognition map of ocean internal wave fringes. a Stripe detection diagram; b stripe recognition diagram

Full size image
Fig. 14
figure 14

Extraction of latitude and longitude of ocean internal wave stripes

Full size image

In order to quantify the detection effect of the algorithm in this paper, the cosine function is used to fit the distribution of fringes, and the Pearson coefficient is used to analyze the correlation of all fitted samples, and the detection effect of the algorithm is quantified through statistical correlation coefficients (Grissa et al. 2011). Obtain the neighborhood pixel value distribution of each pixel of the contour. That is, for each pixel of the contour in Fig. 14, according to its location, the pixel value distribution around the pixel is extracted from the original image data (Hachicha and Job 1994). The rectangular window is used for extraction, that is, the pixel is the central pixel, the size of the rectangular window is determined, and the pixel value distribution is obtained (as shown in Fig. 15).

Fig. 15
figure 15

Schematic diagram of extracting contour neighborhood pixels point by point

Full size image

Detection results of ocean filtering by remote sensing image boundary characteristics

This article elaborates on what happened at 14:11 on August 15, 2006, at the center latitude and longitude 21° 11′ N, 116° 45′ E (20° 40 ' –21° 47 ' N, 116° 10–117° 10 ' E). The filtering method and related image processing algorithm of ocean internal wave SAR image to better detect ocean internal wave. The detection algorithm in this paper can not only remove the contours of non-internal wave stripes, but also remove some small and inconspicuous contours of internal wave stripes and extract the latitude and longitude information of each stripe (Hachicha 1998). At the same time, the algorithm can calculate the position of the light and dark stripes and the distance between the light and dark stripes according to the actual position of the center pixel and can obtain the approximate propagation direction of the ocean internal waves by fitting a straight line to the contour of the internal wave stripes (Hachicha 2007). Finally, the Pearson coefficient is used to quantify the recognition effect of ocean internal wave fringes.

In order to further test the robustness and stability of the algorithm, this paper uses the above algorithm to process another 5 SAR internal wave images (Hartkamp et al. 1999). The data information is shown in Table 5.

Table 5 Marine SAR image information table
Full size table

The above processing is performed on the five marine internal wave SAR images in Table 5, and the processing results are shown in Fig. 16. The statistical distribution of Pearson coefficients corresponding to each image is shown in Table 6.

Fig. 16
figure 16

Internal wave A fringe recognition diagram. a Column separation neighborhood processing image; b Canny processing image; c fringe detection map; d fringe recognition map; e light and dark fringe fitting map; f light and dark fringe extraction map; g Pearson coefficient distribution map; h stripe latitude and longitude extraction map

Full size image
Table 6 Sample Pearson coefficient statistics table
Full size table

In summary, the algorithm in this paper can better recognize images with strong internal wave information and has better recognition accuracy, robustness, and stability. Not only can the contours of the ocean internal waves be recognized, but also the propagation direction of the ocean internal waves can be obtained (Holah 2005). At the same time, since the pixel coordinates and the latitude and longitude coordinates are one-to-one correspondence, the specific positions of the light and dark stripes and the spacing between the light and dark stripes can also be determined (Huang et al. 2014).

Discussion

A preliminary understanding of Japanese vocabulary learning strategies

Classification learning according to Japanese vocabulary semantics

For students majoring in Japanese, they must first master some Japanese vocabulary when they are just getting in touch with Japanese. When learning Japanese vocabulary, you can classify and learn according to the semantics of Japanese vocabulary (Huang et al. 2016). For example, divide the vocabulary into nouns, adjectives, etc., and then sort the similar vocabulary, such as according to the pronunciation of Japanese or the writing characteristics of Japanese, so that the Japanese vocabulary will be classified at multiple levels, which is convenient for students to master and learn Japanese; it also provides basic guarantee for students to review and consolidate vocabulary in the future, and through the mastery of Japanese semantics, it enriches students’ Japanese vocabulary, laying a solid foundation for them to learn Japanese sentences and build a Japanese environment in the future.

Learn Japanese vocabulary with the help of a dictionary and guess the meaning of words

The method of looking up the dictionary can also improve the students’ Japanese vocabulary learning ability, but there are individual differences among some students, and the learning effect produced by this method will vary from person to person. Because the dictionary can only be used as an auxiliary learning tool, certain query methods must be mastered in the process of Japanese vocabulary learning. For example, students can learn the semantics and usage of new vocabulary by looking up the dictionary. In Japanese language learning, students can also learn by guessing the meaning of Japanese words. This method requires guessing the meaning of words according to the Japanese context. In addition, students must master the knowledge and use structure of related Japanese vocabulary, such as common word formation and the nature of the vocabulary to guess the meaning of the word. For the guessing method to learn Japanese vocabulary, sometimes, it will be inaccurate, but for some compound vocabulary and acronyms, through structure division and analysis, and then guessing, the meaning of the word is one of the effective ways. Students can effectively combine the above two methods and use the dictionary to query when the guessing method is inaccurate.

Learning and memorizing Japanese vocabulary through imagination

Imagination is that students learn Japanese vocabulary through imagination. For example, in language learning, there are synonyms, antonyms, or phonetic words, etc. Therefore, in Japanese vocabulary learning, vocabulary can also be imagined, and a vocabulary can be memorized first, and then based on its synonym, antonym, or phonetic features, imagine other words. Through this learning process, imagination storage space is established in student’s brain, thereby enhancing student’s vocabulary memory ability. In imaginative learning, students can not only imagine a single vocabulary, but also imagine a combination of multiple vocabulary. When using the imagination method, students must use certain association rules and explore the relevance of vocabulary and the uniqueness of individual words, so as to stimulate students’ imagination and thinking in Japanese use, which not only increases students’ vocabulary reserves, and it is convenient for students to remember vocabulary. In the process of Japanese vocabulary part-of-speech learning, imaginative vocabulary learning should also be carried out in conjunction with the characteristics of the correspondence between the vocabulary in Japanese sentences and the application of grammar.

Guide students to develop a Japanese vocabulary learning plan

In the teaching of Japanese vocabulary, teachers must first allow students to set up several stages of vocabulary learning and formulate specific learning plans in terms of vocabulary listening, writing, reading, and use. At the same time, the learning effect is evaluated through related evaluation methods. For example, when students first come into contact with Japanese, they can start from the rules of Japanese writing and pronunciation, first understand and analyze the characteristics of Japanese vocabulary, so as to lay a solid foundation for the subsequent study plan.

Teaching strategies of Japanese trade vocabulary

Constructing a student-based classroom teaching model

In Japanese vocabulary teaching, teachers should pay attention to building student-based classrooms with students as the main body of teaching. Teachers should transform from traditional theoretical teaching to the teaching philosophy of cultivating comprehensive Japanese professionals, which is conducive to cultivating outstanding Japanese practical talents. This model takes students as the main body. According to the actual situation of the students, teachers carry out teaching activities with the goal of cultivating students as practical Japanese talents in the teaching curriculum and content. In this process, students have always occupied a dominant position and become constructor of the teaching system in the teaching process. Through the creation of the student-based classroom teaching model, the core teaching concept of teachers can be changed, and teachers will be promoted to innovate and reform the construction of teaching content and classroom environment, actively create a classroom atmosphere that stimulates students to learn independently, and at the same time build an image of student-based classrooms, such as the construction of the Japanese context, can meet the needs of social development for Japanese practical talents.

Infiltrate Japanese national culture into teaching

Teachers should also pay attention to the infiltration of Japanese national culture in Japanese vocabulary teaching. In this process, teachers should select topics carefully to stimulate students’ perception of Japanese national culture, so that the learning of Japanese vocabulary can be integrated into the diverse Japanese national culture. At the same time, language is also an important component of national culture. In fact, learning of Japanese vocabulary, it is also a study of Japanese national culture. Therefore, the teaching of Japanese vocabulary should be closely integrated with the Japanese national culture. Teachers can not only allow students to master the teaching content of Japanese vocabulary, but also take a variety of ways to let students contact and understand Japanese national culture.

When learning a language, its basic vocabulary is the key to learning, and mastering the basic language vocabulary is of great significance to language learning. The learning of Japanese vocabulary is a complicated process. Through the application of a variety of learning strategies during the learning process, students can master and consolidate Japanese vocabulary proficiently. Teachers should infiltrate Japanese national culture from the construction of student-based classrooms and teaching, vocabulary teaching, while combining the actual situation of the students to set up a reasonable course.

Talent training plan based on the specific situation of Japanese trade

Optimize the curriculum system and develop the talent training model of “language + specialty” in an all-round way

From the survey, the Japanese test with the highest recognition is still the “JPLT,” which is generally required to be at level N2 or above, while translation posts require language proficiency to reach the level of N1. In addition to certificates, companies value the students’ comprehensive language skills such as listening, speaking, writing, and translation skills. Therefore, when formulating the talent training plan, we still ensure sufficient class hours for basic courses such as Japanese intensive reading, Japanese listening, and Japanese conversation to consolidate the students’ language foundation.

According to the survey, the most well-known Japanese test is still “JPLT,” which generally requires level N2 or higher, but the translation position requires level N1. In addition to the certificate, the company also highly evaluates students’ comprehensive language skills such as Japanese listening and speaking skills, writing skills, and translation skills. Therefore, in the intensive Japanese reading, Japanese listening, Japanese conversation, and other basic courses, sufficient teaching time is guaranteed to lay a solid language foundation for students’ Japanese learning. Japanese majors study language with work as the center, mainly for the trade market. This requires that at the beginning of the sophomore year, in addition not only to reading more business Japanese conversation, listening business Japanese translation, and Japanese business writing courses, but also constantly getting involved in international trade, foreign trade, foreign trade communication documents, and setting up international marketing and other majors, the course can fill in the language weaknesses and lack of professional knowledge of students in ACTS. In addition, it can be deeply understood from the survey that “culture lags behind business.” Japanese cultural knowledge and Japanese business etiquette can be a lubricant for good communication with Japanese customers and ensure the smooth completion of orders. Therefore, the country profile of Japan, Japan’s geography, history, business etiquette, and Japanese business culture all need to be mastered.

School-enterprise cooperation to jointly build a team of high-quality “double-qualified” teachers

Teacher’s level directly determines the teaching effect. Building a team of high-quality teachers has a great influence on the training of applied talents in colleges and universities. The current professional teachers have relevant professional qualifications such as translation certificate, international clerk certificate, and economist certificate. Due to the lack of experience in trading companies, the school will arrange for them to intern in trading companies for half a year. According to the needs of future education, they will sign specific internship contracts to ensure that teachers can master relevant skills in the company. The trade industry will change a lot at any time. Therefore, teachers’ internships are rotated every 5 years so that teachers can update their knowledge in time. In addition to the long-term training in the company, it is recommended that teachers take short-term study in the company and use the winter and summer vacations to practice in the company for 3 months.

In addition to vigorously training excellent full-time teachers, the school also hired management staff from trading companies as part-time teachers of the school, responsible for Japanese business training, on-site training, conference training, etc. According to the operation process of students in vocational skill training, students can experience the goals of “learning through work” and “working through study.” In addition, business personnel are regularly invited to give lectures to students to improve students’ practical ability, broaden their horizons, and help students understand the development trends of the industry and the company as quickly as possible, clarify the professional knowledge required by the company, and formulate their own career plans. In the internship stage, in order to achieve seamless connection between classrooms and employment positions, the school needs to hire industry experts as practical education instructors.

Deepen school-enterprise cooperation and strengthen practical teaching

According to the national medium- and long-term education reform and development plan, the Ministry of Education has led some opinions on the scientific development of vocational education with regard to the promotion of higher vocational education reform and innovation. The guidance of the Ministry of Education on promoting the reform and development of vocational education indicates the combination of school construction and the actual situation of development, by motivating college employees to actively participate in the development of cooperation projects between universities and enterprises, and real-time tracking feedback, expanding educational resources, so as to cultivate application talents with excellent professional qualities.

Regarding on-campus practice, on the basis of the original school, a language training center, a foreign trade exhibition, and other on-campus training bases for the application of Japanese have been established. At the same time, the base is equipped with specialized laboratory personnel responsible for management. Refer to the management rules of the enterprise, adopt a standardized practical training management system, clarify responsibilities, and carry out work. Taking into account the characteristics of the zero-start Japanese major, the configuration of the internship courses on campus, from basic language skills to the integration of language and business skills, is graded from simple to complex. According to the actual position and industry of the students, specific training courses such as administrative internships, business internships, conference internships, and foreign-related exhibitions internships are set up. The training is to arouse the enthusiasm of students and is combined with research on business Japanese teaching materials.

Regarding off-campus internships, schools and companies cooperate, according to the actual needs of both parties, so as to build new application talents suitable for development. In the process of cooperation, companies can truly feel the effects of school teaching, realize the needs of companies, schools, and students, and finally achieve the goals of cooperation. For example, when students are selected to participate in off-campus training, students can experience actual work scenarios to meet the short-term talent needs of enterprises. In addition, the construction of the “Trade Show Service Center” is promoted in the school. Interested students learn knowledge related to the work flow of foreign trade e-commerce through internships at the school. At the same time, as the “zero connection” between the school and the enterprise, it also create value for the enterprise.

Two-pronged approach of “hardware” training and “software” training

Regarding the solid foundation of Japanese and excellent professional skills as the hardware of talents, qualities such as teamwork and diligence must be regarded as software. According to the survey, in addition to language skills, professional skills, and certificates, companies also focus on professional qualities such as expressiveness and communication skills, loyalty, teamwork, self-discipline learning ability, innovation ability, and anti-pressure ability when recruiting. Therefore, in terms of cultivating talents, schools cannot ignore the quality of talents.

Comprehensively promoting high-quality education for students and clarifying student training goals and the comprehensive quality of students can be evaluated scientifically and reasonably. Based on the actual situation of the school, students’ extracurricular activities have been formulated. The content of cultivating students’ comprehensive quality in the school is divided into six projects including participating in corporate internships, participating in corporate projects, participating in competitive activities, obtaining professional qualification certificates, participating in social practice activities, and participating in forum speeches. In addition, as the basis for evaluating students’ excellent comprehensive quality, the important experience and main results of students in school learning are recorded in detail, so as to continuously improve the comprehensive quality of students themselves. In addition to the guidance of this system, teachers’ integration of professional abilities and professional ethics into the curriculum during the education process is also very important for talent training.

Conclusion

According to the “Marine Inspection” proposal, islands are very important for China’s territorial protection and the maintenance of maritime rights. In this paper, based on the analysis of existing image segmentation algorithms, considering the large-area observation and multi-band characteristics of remote sensing images, an ocean detection algorithm based on the boundary characteristics of remote sensing images is proposed. Analyzing the neighborhood of all pixels of each contour uses cosine function fitting and Pearson coefficient to demonstrate the effectiveness and accuracy of the method in this paper pixel by pixel. Analyzing multiple SAR images of ocean internal waves verifies the robustness and stability of the method and lays the foundation for the next step in the identification of ocean internal waves based on neural networks and other intelligent algorithms. The research results show that for images with strong internal wave information, this method can better identify the light and dark fringes of ocean internal waves and not only can effectively remove the contours of non-internal wave fringes and some small and inconspicuous internal wave fringes, and the identification of fringes can be verified pixel by pixel by cosine function fitting. For images with weaker internal wave information, speckle noise will have a more significant impact on the detection effect of the Canny operator, resulting in a decline in the detection effect, thereby affecting further recognition. In the learning of Japanese vocabulary, it is necessary to focus on the relevant learning strategies, formulate a vocabulary learning plan that meets the characteristics of their own learning, and constantly create a Japanese context during learning to improve the comprehensive use of vocabulary. As a teacher, we must use Japanese vocabulary learning strategies in teaching to cultivate students’ independent learning ability, help students build vocabulary knowledge structure, and combine with teachers’ vocabulary teaching methods to effectively improve students’ ability to learn Japanese and enable students in gaining a richer Japanese vocabulary reserve which has laid a good foundation for his future Japanese learning and development.